WO2015093330A1 - 認識データ伝送装置 - Google Patents
認識データ伝送装置 Download PDFInfo
- Publication number
- WO2015093330A1 WO2015093330A1 PCT/JP2014/082371 JP2014082371W WO2015093330A1 WO 2015093330 A1 WO2015093330 A1 WO 2015093330A1 JP 2014082371 W JP2014082371 W JP 2014082371W WO 2015093330 A1 WO2015093330 A1 WO 2015093330A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- information
- person
- action
- data
- recognition
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B13/00—Burglar, theft or intruder alarms
- G08B13/18—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength
- G08B13/189—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems
- G08B13/194—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems
- G08B13/196—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems using television cameras
- G08B13/19602—Image analysis to detect motion of the intruder, e.g. by frame subtraction
- G08B13/19613—Recognition of a predetermined image pattern or behaviour pattern indicating theft or intrusion
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B13/00—Burglar, theft or intruder alarms
- G08B13/18—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength
- G08B13/189—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems
- G08B13/194—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems
- G08B13/196—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems using television cameras
- G08B13/19663—Surveillance related processing done local to the camera
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N7/00—Television systems
- H04N7/18—Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast
- H04N7/188—Capturing isolated or intermittent images triggered by the occurrence of a predetermined event, e.g. an object reaching a predetermined position
Definitions
- the present invention relates to a recognition data transmission technique.
- Equipment and methods have been proposed for recognizing human actions using various sensors, such as the presence of people, room entry / exit management, and remote monitoring.
- Installation can be performed by transmitting a home image captured by the camera to the server device via the telephone line and the Internet, and receiving control commands including the image and camera shooting operation from the server device to the mobile terminal and performing control.
- An easy and inexpensive remote monitoring system has been proposed (see Patent Document 1 below).
- a face included in the video content is detected, a feature amount related to the face is recorded in the content management file, and the feature amount related to the specific face is compared with the feature amount recorded in the content management file to include the specific face
- the person detection system described in Patent Document 1 starts photographing a person when a person is detected by a human body sensor, and transmits the image to a transmission destination.
- the watching side can always watch the action of the person, but the watcher himself / herself needs to determine whether the person is taking a dangerous action or spending it safely. Therefore, there is a problem that the watching side needs to always watch the person of the transmitted video.
- the transmission data is video data, there is a problem that the privacy of the person photographed by the camera cannot be protected. Furthermore, the amount of data to be transferred has become enormous, and it has been necessary to increase the data transmission band.
- Patent Document 2 can record and search a face detected from video content and its feature amount, but cannot detect a person's action in the video and record the information. Furthermore, since the person detected from the video content and its feature data are recorded for each representative thumbnail created at regular intervals, the detection data of the same person is duplicated and the detection information increases. was there. In addition, since multiple representative thumbnails may be recorded within one action time of the detected person, the relationship between the recorded detection information is unknown and the person's action could not be linked .
- the present invention has been made in view of such problems, and detects the person detected from the video and its action, and transmits the information in units of the person's action, thereby improving convenience and transmission on the watching side. It is an object of the present invention to provide a recognition data transmission apparatus that achieves both data reduction.
- At least one imaging unit, and a person detection unit that detects a feature amount of a person from an image captured by the imaging unit and identifies a person in the image from the detected feature amount;
- the user information detecting unit for detecting user information representing the physical characteristics of the person from the feature amount related to the person detected by the person detecting unit, the user information output by the user information detecting unit, and the imaging unit
- a motion detection unit that detects a user's motion (motion or gesture gesture) from an image, a motion recognition unit that recognizes a behavior including a person's motion from the motion information output by the motion detection unit and the user information
- a data conversion unit that divides the recognition information output by the action recognition unit on a time axis for each human action, and generates divided block data, and the data conversion There the block data to be output, and a transmission control section for controlling to transmit to the outside for each block, recognition data transmission apparatus characterized by having a is provided as a recognition data.
- user information is detected by comparing with data in an information DB registered in advance as a feature amount of a person.
- the behavior is detected by comparing with data in the information DB registered in advance as motion information.
- the present invention by detecting the person detected from the video and the action thereof and transmitting the information in the action unit of the person, it is possible to achieve both the convenience of watching and the reduction of the transmission data.
- the present invention by detecting the person detected from the video and its action and transmitting the information in units of the person's action, it is possible to achieve both the convenience of watching and the reduction of transmission data.
- FIG. 1 It is a figure which shows an example of the information which shows the video data corresponding to the action information at the time of abnormality in the recognition data transmission apparatus by the 3rd Embodiment of this invention. It is a figure which shows the example of a video data structure transmitted with the block in the recognition data transmission apparatus by the 3rd Embodiment of this invention. It is a figure which shows the kind of codec in the recognition data transmission apparatus by the 3rd Embodiment of this invention. It is a flowchart figure which shows the flow of the video data transmission process corresponding to the action information at the time of abnormality by the 3rd Embodiment of this invention.
- FIG. 1 is a conceptual diagram of a monitoring system using a recognition data transmission apparatus according to the first embodiment of the present invention.
- the recognition data transmission device 1 detects the feature data and behavior information of the room occupant 7 detected from the imaging device 10 and the captured video, and transmits the behavior information to the outside.
- FIG. 1 shows an example in which the recognition data transmission device 1 including the imaging device 10 is installed at the upper part of the television 3 as an example of the installation position.
- the television 3 is installed at a corner of a room so that the entire room 5 can be photographed from the recognition data imaging device 10. Moreover, it installs in the corner on the opposite side of the door 6 so that the occupant 7 entering the room through the door 6 can be surely photographed.
- the recognition data transmission device 1 captures the person in the room 7, detects a person from the captured video, information on who the person is, and what action he is taking (hereinafter referred to as “behavior information”). ) Is detected.
- the installation position of the recognition data transmission device 1 may be anywhere as long as the entire room 5 can be looked over and the room occupant 7 is surely within the angle of view of the captured video. is not.
- it may be inside a ceiling, lighting, or inside an air conditioner installed on an indoor wall surface.
- the behavior information detected by the recognition data transmission device 1 is stored in a separate room of the same building as the room 5 using local network (LAN) means such as Wi-Fi, Ethernet (registered trademark), Bluetooth (registered trademark) or the like.
- LAN local network
- the information is displayed on the information display terminal 2.
- the detected action information may be displayed as characters or illustrations, or may be converted into voice data by an information display terminal and output as voice data.
- the action information may be output to the information display terminal 2 together with a flash, a warning sound, a vibration, or the like to notify the watcher.
- the data transmitted to the information display terminal 2 is action information
- an image captured by the imaging device 10 of the recognition data transmission device 1 may be transmitted.
- the watcher can instantly confirm the condition and safety of the room occupant 7 by transmitting video data together with action information.
- the action data or captured video detected by the recognition data transmission device 1 is connected to the Internet using an optical line, ADSL, or a wireless communication system for mobile phones (3rd and 4th generation mobile communication systems), and remotely. It may be transmitted to a certain information display terminal 4. As a result, it is possible to notify the watcher in the remote place of the watch information output to the information display terminal 2.
- FIG. 2 is a diagram illustrating a transmission example of behavior information detected by the recognition data transmission device 1. The case where the persons whose behavior is detected is A and B is illustrated.
- the passage of time is shown from the left to the right in FIG. This represents that the person A first performed the “entering” action, and after the “entering” action was finished, the “standing” action was performed after a lapse of a certain time.
- the left and right arrows on the drawing represent the elapsed time required for the corresponding action.
- the actions of the persons A and B are shown in time series together with the elapsed time.
- a person's action rarely occurs only at a certain moment, and generally there is an elapsed time from the start to the end of the action.
- the action information may be notified for each action. Therefore, for example, the behavior information regarding the behavior of “standing” may be transmitted when the behavior is started and how long the elapsed time is.
- the behavior information is transmitted in units of frames, since the relevance of the behavior information detected between frames is unknown, it is necessary to add information that holds the relationship, and transmission data increases.
- this embodiment is characterized in that one action is transmitted from the start to the end of the action as the minimum unit of action information.
- the behavior information and the video data may be referred to by the start time and the end time of the behavior, so that the video corresponding to the behavior can be searched.
- FIG. 2 shows that the person A first starts the “entering” action and the person B starts the “sitting” action during the “entering” action.
- the behavior information related to the first person A's “entering” behavior is stored in block 0 and transmitted.
- the action information “sitting” performed by the person B is set as a block 1 and transmitted after the block 0.
- the detected action of the person is detected as one action information for each action, and is transmitted as a block in the order in which the actions occurred.
- a person's action takes an elapsed time for one action
- another action by another person may be performed while a certain action is being performed. For example, after the person B in FIG. 2 starts “crying”, the person A starts “standing” and before the person B “crying” ends, the action of the person A “standing” ends. This is the case.
- the transmission of the behavior information is started when the behavior of the person B “crying” is started, the behavior information of the person A “standing” cannot be transmitted until the behavior of “crying” ends. End up.
- FIG. 2 shows an example in which the action information “block is lost” of the person A whose action has been started is transmitted as the block n before the action information “block n + 1” of the person A.
- the priority of action information may be set higher in order from the earliest action start time, and the action information may be transmitted in descending order of priority. If there is action information that ends at the same time and the action information also has the same action start time, a person with a high priority is set in advance, and a higher priority is given to the action of a person with a higher priority.
- the behavior information may be transmitted in descending order of priority. In FIG. 2, it is set that the priority of the person A is higher than that of the person B, and the person A “crying” and the person B “wake up” have the same start time and end time. In this example, “crying” of the person A with high priority is transmitted as a block 6 before the block 7 of “wakes up” of the person B.
- the behavior information described above has been described as an example in which the behavior information is transmitted in units of blocks as information representing the behavior of a person. However, not only the behavior information but also a corresponding video or still image may be transmitted as data to be transmitted. Further, only action information may be transmitted in order to reduce the amount of transmission data.
- recognition data transmission device A configuration example of the recognition data transmission apparatus according to the present embodiment will be described with reference to FIG.
- the recognition data transmission device 1 generates a video signal representing the video captured by the imaging device 10 and sends the generated video signal to the video analysis unit 11 as video data.
- the imaging device 10 may be composed of one or more imaging units (101 to n). In FIG. 3, it is illustrated that there are n imaging units. By using one or more imaging units, it is possible to divide and shoot the indoor area captured by each imaging unit, so that it is possible to capture a wider area, or an area that becomes a shadow of objects such as furniture (Occlusion area) can be reduced.
- the imaging units 101 to n are, for example, a camera including an optical system including a lens that focuses light incident from a subject on a focal point and an imaging element that converts the collected light into an electrical signal.
- the imaging devices included in the imaging units 101 to n are, for example, a CCD (Charge Coupled Device) element and a CMOS (Complementary Metal Oxide Semiconductor) element.
- the output of the video imaged by the imaging device 10 may be wired or wireless, and may be input to the video analysis unit 11 using any transmission method.
- the video analysis unit 11 analyzes the video input from the imaging device 10 to detect or identify a face or a person, or detect an action taken by the person.
- the information DB (memory) 12 stores learning information such as a face or object template necessary for detecting or identifying a face or a person, a motion information template necessary for detecting an action, or the like.
- intermediate data for video analysis is temporarily stored.
- feature amount data of the face and person is stored.
- information for displaying an action for a person's action on the information display terminal on the watching side of the network destination according to the recognition information detected by the video analysis unit 11 is stored.
- the video analysis unit 11 detects an action of a person in the video using the input video of the imaging device 10 and the learning data of the information DB 12 and outputs it as recognition information. At that time, the video data (including the captured video captured by the imaging device 10) used for the action detection may be output together with the corresponding recognition information.
- the data conversion unit 13 performs processing for converting the recognition information and the captured video output from the video analysis unit 11 into the above-described block units.
- the recognition information and video data converted into blocks are sent to the transmission control unit 14 as behavior information, and the transmission control unit 14 transmits the information to the outside using a transmission unit such as Wi-Fi.
- the transmission control unit 14 may not only transmit to a network but also record it on a recording medium 16 such as an optical disk, a semiconductor memory, or an HDD.
- the behavior information transmitted to the outside by the transmission control unit 14 may be configured only by recognition information, may be configured only by video data, or may be configured by both behavior information and recognition information. Also good.
- the video input to the video analysis unit 11 may be a video database video recorded in the recording medium 16 as well as a video shot from the imaging device 10.
- the video database may be a video stored in a server connected to the Internet.
- the video data input to the video analysis unit 11 may be not only video data but also detection results from various sensors such as an infrared sensor, a distance measuring sensor, an acceleration sensor, a gyro sensor, and an electronic compass, and the detection results of these sensors are used. Recognition information may be generated.
- An infrared sensor a distance measuring sensor
- an acceleration sensor a gyro sensor
- an electronic compass an electronic compass
- the video analysis unit 11 combines the detection of a person's face based on the video signal input from the imaging device 10 and the motion information calculated from the difference in video data having a plurality of shooting times to detect the face and the area on the detected video From this, it is determined that the moving object (the area where the image data has changed) is a person, whether or not there is a person, and tracking the person is detected, and the result is output as recognition information.
- the video analysis unit 11 detects the position of the person and the amount of motion based on the video signal input from the imaging device 10, identifies the gender and age of the person, who the person is, and the direction in which the face is facing It detects what kind of behavior the detected person is doing, such as a facial expression and a facial expression, and outputs the result as detection information. Furthermore, information representing human motion such as gestures is detected and output as the action to be detected.
- the motion vector detection unit 110 detects a moved area in the video from a plurality of videos input from the imaging device 10.
- the motion vector detection unit 110 inputs the detected moved area and its center-of-gravity coordinates, movement amount and movement direction (vector) as tracking information (motion vector information) to the person detection unit 111 and the motion detection unit 113, and simultaneously detects them. Output as information.
- the motion vector detection unit 110 may input person information output by a person detection unit 111 described later to detect a moved area in the video ( No illustration). It is determined that a person has moved when the moved area detected by the motion vector detection unit 110 is within a predetermined threshold range (area) based on position information included in the person information output by the person detection unit 111. Then, the detected moving area, its center-of-gravity coordinates, movement amount, and movement direction (vector) may be input to the motion detection unit 113 as person tracking information (motion vector information) and simultaneously output as detection information. In this way, by combining the detected moving area information and person information, it is possible to avoid detecting motion vector information other than a person such as an object, and extracting motion vector information of a person to be watched, It becomes possible to output.
- the motion information detection method of the motion vector detection unit 110 is not limited to this method, and the current frame in units of areas (macroblocks) such as 8 ⁇ 8 or 16 ⁇ 16 pixels in the vertical and horizontal directions. Any method may be used, such as a method of searching where in the past frame a region matching the macroblock is found, and if the region is found, the movement amount of the coordinate is used as the movement amount.
- the person detection unit 111 detects a region representing a facial image from a plurality of videos input from the imaging device 10 and motion vector information input from the motion vector detection unit 110.
- the person detection unit 111 generates two-dimensional face area information representing two-dimensional coordinates of representative points (for example, centroid points) in the detected face area and two-dimensional coordinates of the upper end, lower end, left end, and right end of the area. .
- the person detection unit 111 combines the above-described two-dimensional coordinates generated from a plurality of images taken by a plurality of imaging units, converts the two-dimensional coordinates into the three-dimensional coordinates of the indoor three-dimensional space coordinate system, and converts the three-dimensional face position information. Generate. If a face is detected here, it is determined that a person has entered the room.
- the person detecting unit 111 extracts, for example, pixels within a color signal value range representing a preset face color (for example, skin color) from the input image signal.
- the person detection unit 111 may include a storage unit that previously stores a grayscale (monochrome) image signal representing a human face. Therefore, the person detection unit 111 calculates a correlation value between the grayscale image signal read from the storage unit and the input image signal for each image block including a plurality of pixels, and the calculated correlation value is greater than a predetermined threshold value. A large image block is detected as a facial region.
- a grayscale (monochrome) image signal representing a human face. Therefore, the person detection unit 111 calculates a correlation value between the grayscale image signal read from the storage unit and the input image signal for each image block including a plurality of pixels, and the calculated correlation value is greater than a predetermined threshold value. A large image block is detected as a facial region.
- the person detection unit 111 calculates a feature amount (for example, Haar-Like feature amount) based on the input image signal, and performs a predetermined process (for example, an Adaboost algorithm) based on the calculated feature amount.
- the face area may be detected.
- the method of detecting the face area by the person detection unit 111 is not limited to the above-described method, and any method may be used as long as it detects the face area from the input image signal.
- a person may be detected by combining facial area information and a motion vector. It is possible to reduce the amount of calculation by tracking the detected person while collating the motion vector information and the facial area information, improving the person detection accuracy in the next frame, and specifying the face detection range.
- the method for detecting a face on an image has been described.
- information about other faces such as gender of the face, age, direction of the face, laughing or angry, crying
- a feature amount is extracted from these images, such as facial expressions such as being present, and registered to detect a face on the image.
- the orientation of the face is detected by extracting and registering the feature quantity from the face image facing in various directions such as up, down, and sideways.
- the person detecting unit 111 extracts a feature amount from a face image of a specific person in advance, registers the feature amount, and collates the input face image with the registered feature amount using the algorithm. It may be possible to identify who the registered face image is.
- Information relating to these faces detected by the person detection unit 111 is input to the user information detection unit 112, the motion detection unit 113, and the action recognition unit 114 as person information, and is simultaneously output as detection information.
- the person detection unit 111 cuts out a captured image captured by the imaging device 10 or a face area or a person area corresponding to the person information detected from the captured image.
- a video corresponding to the person information may be output to the user information detection unit 112, the motion detection unit 113, and the behavior recognition unit 114.
- the attribute estimation unit 1120 estimates the age and gender of the detected person information in comparison with the data in the information DB registered in advance as the feature amount of the person.
- the facial expression estimation unit 1121 estimates facial expressions such as laughing, crying, and anger of a person.
- the person identifying unit 1122 compares the feature quantity of the facial organs of the person registered in advance in the information DB, collates with the registered person, and identifies the person in the video data. If it matches the registrant, the person's name is output. If no one matches, the identification result is output as an unknown person.
- the position detection unit 1123 outputs the position of the person on the video.
- the human body part detection unit 1124 detects a part of the human body such as the detected head, arm, hand, foot, and torso of the detected person, and outputs position information and size in the video.
- the motion detection unit 113 is detected from a plurality of videos input from the imaging device 10, person information input from the person detection unit 111, and tracking information (motion vector information) input from the motion vector detection unit 110. Detect motions such as human movements and actions.
- the coordinate position in the three-dimensional spatial coordinate system of the face input from the person detection unit 111 is higher than a predetermined threshold, it is determined that the user is standing.
- the height information is registered together with the identification information of the person registered in advance in the information DB, and the three-dimensional coordinates of the face determined to be standing from the identification information and the height information output from the person detection unit 111 If the range is set, it can be determined that the player is standing with higher accuracy.
- the motion detection unit 113 outputs these determination results as motion information.
- the motion detection unit 113 includes the person information from the person detection unit 111, the tracking information of the motion vector detection unit 110, the person position information of the position detection unit 1123, and the human body part information of the human body part detection unit 1124.
- the person information from the person detection unit 111 When it is detected that a person exists in the room and is moving, it is determined that the detected person is awake. Further, when the person who is awake suddenly falls asleep or does not move for a while after sleeping, a sudden change in physical condition is predicted, so information that the person is falling is output as motion information.
- the motion detection unit 113 includes the person information from the person detection unit 111, the tracking information of the motion vector detection unit 110, the person position information of the position detection unit 1123, and the human body part information of the human body part detection unit 1124.
- a jump or the like is detected from the movement of a person's hand or foot or the change in the three-dimensional coordinates of the face, and a gesture such as a predetermined body shape or movement is detected. Then, the detected gesture information is output as motion information.
- the behavior recognition unit 114 detects the presence / absence of a person in the room, which is detection information input from the user information detection unit 112, the identification result of the person who is in the room, the motion detection result of the occupant
- the behavior of the room occupant is detected from the information in the information DB 12 and recorded in the information DB 12.
- the detected behavior includes detecting whether the occupant is standing, sitting, sleeping, where in the room, and in which direction it is facing. Further, by detecting the position where the person is in and the direction in which the person is facing, for example, watching TV, cooking, eating, etc. can be detected.
- the detection result may be displayed on a display such as a television. It may display current detection information in real time or display past data.
- the action recognition unit 114 may transmit the detected action information to the watching side through the network.
- the behavior information to be transmitted may be transmitted / displayed or recorded on a display display (TV, mobile phone, portable terminal, dedicated display, etc.) on the side of watching.
- the action information detected by the server existing on the network may be recorded once, read when necessary, and displayed on the display.
- the action recognition unit 114 can prevent confinement and drowning accidents by detecting entry into a bathroom or a washing machine installation room, for example, in the case of a child from detection information of a person who has entered the room. Or it can prevent fires due to stove and stove mischief in the kitchen, and can also be used for indoor security monitoring. Further, it is possible to detect a thief or the like by detecting entry of a person who is not registered in the information DB 12.
- the action recognition unit 114 automatically recognizes a predetermined gesture or a predetermined person's action, and operates the corresponding devices 1 to m.
- the detection information input from the user information detection unit 112 and the information in the information DB 12 are collated, a predetermined device designation and a control command for the designated device are searched, and the control command is transmitted to the corresponding device.
- the action recognition unit 114 is a face corresponding to a photographed video captured by the imaging device 10 or personal information detected from the photographed video input from the person detection unit 111, the user information detection unit 112, and the motion detection unit 113.
- the data conversion unit 13 You may output to information DB12.
- FIG. 5 is a diagram illustrating the order and contents of data stored from the top of one block.
- the data stored in the block is binary format data according to the format of FIG.
- Each piece of information (hereinafter referred to as a field) stored in the block is recorded in order of block ID, block priority, action ID, start time, action time, frame rate ID, additional information size, and additional information from the top.
- Block ID The block ID of the first field is a unique ID added to each block, and is stored in order from 0 for the block to be transmitted.
- the block ID recording area has a fixed length of 3 bytes, and stores a value from 0 to 0xFFFFFF.
- the block priority of the second field stores the priority of the detected behavior information. The higher the priority, the earlier (at an earlier time) the action was performed (finished), and the more important action information that requires an emergency such as “ fallen”.
- the block priority recording area has a fixed length of 1 byte and stores a value of 1 to 0xFF. A value of 1 indicates the lowest priority, and the higher the value, the higher the priority. Also, 0xFF indicates emergency data (highest priority).
- Action ID The action ID in the third field is an ID representing the content of the detected action.
- the action ID recording area has a fixed length of 1 byte, and stores an ID corresponding to the kind of action as shown in the action ID of FIG.
- the action type ID corresponding to the action information recognized by the video analysis unit 11 is stored. For example, in the case of “entering a room” which is the first action of A in FIG. 2, ID2 is stored.
- “no action information” of ID 0 is stored.
- “no action information” is stored and transmitted even though there is no conspicuous action, a person is detected in the captured image of the imaging unit 101, who the person is, and the detected face Or when it is desired to transmit additional information such as reliability of behavior.
- the action ID may be additionally defined later in addition to the actions 0 to 15 shown in FIG. Therefore, the action IDs 16 to 254 are reserved as reserved areas. When adding action information later, action IDs 16 to 254 are used to newly define action types.
- the start time of the fourth field stores information indicating the time when the action indicated by the action ID is started.
- the recording area of the start time has a fixed length of 4 bytes, and stores the frame number from the start of shooting when transmitting video data shot by the imaging unit 101 and action information recognized from the video data. By storing the frame number corresponding to the captured video after the start of watching, the video corresponding to the recognized action can be immediately searched.
- time (time) information may be stored when behavior information is transmitted in order to reduce the amount of data to be transmitted or in the case of a system that does not require video data.
- the action time in the fifth field stores information indicating the elapsed time spent on the action indicated by the action ID.
- the action time recording area has a fixed length of 4 bytes.
- the elapsed time spent on the behavior may be stored.
- the frame rate ID of the sixth field stores information indicating the frame rate of video data captured by the imaging unit 101.
- the recording area of the frame rate ID has a fixed length of 1 byte, and stores an ID corresponding to the frame rate type shown in FIG. For example, if the video data is 60 frames per second, ID value 8 is stored.
- IDs 14 to 255 are reserved as reserved areas so that the frame rate can be added later. If a frame rate is added, the IDs 14 to 255 may be additionally defined.
- the frame rate ID calculates the action start time (time) and the elapsed time of the action from the frame number and the number of frames when the frame number and the number of frames of the video corresponding to the information indicated by the start time and the action time are stored. Use for. For example, if the frame rate ID is 8 (60 frames of video per second) and the start time is 600 (frames), this indicates an action started 10 seconds after the start of shooting. If the action time is 3600 (frames), it indicates that the action has been performed for 60 seconds after the action is started.
- a frame rate ID of 0 indicates that there is no frame rate information. This is used when the recognized action information is transmitted and the corresponding video data is not transmitted. In this case, even if the frame number or the number of frames is stored in the start time or the action time, it cannot be converted into time (time) information. Therefore, the time (time) information is stored in the start time or action time field. . Therefore, when the value of the frame rate ID is 0, it can be used as flag information for determining that the information stored in the start time or action time is time (time) information. (Additional information) The additional information size of the seventh field and the additional information of the eighth field store detection information used to determine the recognition information output from the video analysis unit 11.
- the additional information recording area has a variable length from 0 to 65535 bytes, and stores the size of the additional information to be transmitted in the additional information size.
- the additional information size recording area has a fixed length of 2 bytes, and the size of the additional information size is stored in bytes.
- the additional information stores action reliability, face identification ID, face identification reliability, text information size, and text information from the top.
- action reliability which is the first field of the additional information, stores the reliability (probability of recognition information) for the action recognized by the video analysis unit 11.
- the action reliability recording area has a fixed length of 2 bytes, and stores a value from 0 to 65535. The larger the value, the higher the reliability.
- Face identification ID which is the second field of the additional information, stores an ID for identifying the person who performed the action recognized by the video analysis unit 11.
- the face identification ID recording area has a fixed length of 2 bytes, and stores a value from 0 to 65535.
- a value 0 represents no identification information (no identification person), and a value 1 represents person identification information.
- the identification information is obtained by photographing a face of a person who wants to be identified in advance, and extracting a person's characteristic as a face information by the person identification unit 1122 of the video analysis unit 11 and assigning it to the identification ID, thereby registering the person to be identified. To do. Then, the feature of the person photographed by the person identifying unit 1122 is similarly extracted from the photographed video from the start of watching as face information, and it is identified which ID of the previously registered identification ID matches, and the matching identification ID is If it exists, the ID is stored in the face identification ID. If there is no matching identification ID (a person different from the person registered in advance), 0 is stored in the face identification ID.
- the face identification reliability which is the third field of the additional information, stores the reliability (probability of recognition information) for the person identification ID identified by the person identification unit 1122 of the video analysis unit 11.
- the face identification reliability recording area has a fixed length of 2 bytes and stores a value from 0 to 65535. The larger the value, the higher the reliability.
- the text information size of the fourth field of the additional information and the text information of the fifth field include a description of the transmission block, the name of the face identifier, the notification content corresponding to the attention and warning for the recognized action, etc. Stores character information.
- the text information recording area has a variable length from 0 to 255 bytes, and stores the size of the text information to be transmitted in the text information size.
- the recording area of the text information size has a fixed length of 1 byte, and the size of the text size is stored in byte units (one character unit).
- Ending text information by adding a null character by adding a null character.
- the text information size stores the size of 4 bytes alignment text. The number of characters actually stored by adjusting the size may be determined by searching for Null characters from the beginning of the text information.
- the information stored in the additional information may store other detected or recognized information regardless of these. Therefore, the Reserved area is secured as a variable-length area so that additional information can be added later.
- the number of action types in FIG. 6A is 256 or more, and information such as the action type is stored using the extension flag of ID255.
- the Reserved area has a variable length size, but since the size of the entire additional information is stored in the additional information size, the end position of the Reserved area can be detected.
- the recognition data transmission process according to the present embodiment will be described.
- FIG. 7 is a flowchart showing a flow of recognition data transmission processing according to the present embodiment.
- the video analysis unit 11 acquires video data from the imaging device 10.
- the process proceeds to step S101.
- the motion vector detection unit 110 holds a plurality of frame videos having different shooting times based on video data captured by the imaging device 10 and detects an area where the pixel value has changed on the video as a moving area. , Output the amount of movement.
- the detected movement amount, movement coordinates, and movement vector are output as motion vector information.
- the current frame image is stored in the information DB (memory) 12.
- the process proceeds to step S102.
- Step S102 The person detection unit 111 detects a person in the video data using the video data captured by the imaging device 10 and the motion vector detected in step S101.
- a face is detected from the video data by the above-described method, and if a face is present, it is determined that a person is present. Further, even when a person is present in the video, a face is not necessarily photographed, so it may be determined that a person is in the moved area using the calculated motion vector. Furthermore, a person may be detected by combining face detection information and a motion vector. By comparing the motion vector output in step S101 with the person information detected in step S102, the detected person is tracked, and the amount of calculation is improved by improving the person detection accuracy in the next frame and specifying the face detection range.
- the personal information such as the presence / absence of a person in the video, the position of the person in the video, the size of the face, the direction of the face, the feature amount of the facial organ, and the tracking result of the person is output, and the process proceeds to step S103.
- step S101 and step S102 you may replace the order of a process as mentioned above (it may be step S101 to step S101).
- the attribute estimation unit 1120 determines the age and gender of the detected person information by comparing with the data in the information DB registered in advance as the feature amount of the person. presume.
- the facial expression estimation unit 1121 estimates facial expressions such as laughing, crying, and anger of a person.
- the person identifying unit 1122 compares the feature quantity of the facial organs of the person registered in advance in the information DB, collates with the registered person, and identifies the person in the video data.
- the position detection unit 1123 outputs the position of the person on the video.
- the human body part detection unit 1124 detects a part of the human body such as the detected head, arm, hand, foot, and torso of the detected person, and outputs position information and size in the video.
- the detected user information is output, and the process proceeds to step S104.
- the motion detection unit 113 detects motion information including the movement of the person and gesture gesture from the position information of the person detected from the plurality of frame images having different shooting times, the human body part information, the motion vector information, and the person tracking information. Then, the process proceeds to step S105.
- Step S105 A person such as a person standing, sitting, sleeping, or lying down based on the video data captured by the action recognition unit 114 and the detected person information, user information, and motion information. Recognize information (behavior information) related to the behavior of and output it as recognition information. Thereafter, the process proceeds to step S106.
- the recognition information output from the action recognition process in step S105 is stored in an information DB (memory). This is used to detect motion information, tracking information, and motion information for recognition between a plurality of frames, and stores recognition information detected from current video data. Thereafter, the process proceeds to step S107.
- information DB memory
- Step S107 In order to transmit the recognition information output from step S106, the recognition information is divided for each person's action, and the recognition information for one action is generated as one block data.
- One block data includes an action ID corresponding to the above-described action start / end time and action. Then, the process proceeds to step S108.
- Step S108 With respect to the behavior detected in the input video data, it is determined whether or not there is other behavior information that has detected that the behavior has ended in the video data, that is, has ended at the same time. If there is action information whose action has ended at the same time (step S108 Yes), the process proceeds to step S109.
- Step S109 When there is action information that ends at the same time, values with higher priorities are stored in the block priority field of the block that stores the action information in order from the earliest action start time. For example, 0x80 (normal data) is stored in the block priority for the action information with the latest action start time. Next, for action information with the earlier action start time, 0x81 is stored in the block priority, and the action information with the earlier action start time stores a higher priority value in the block priority. .
- step S110 The behavior information detected from the video data is transmitted in units of blocks. As for the transmission order, transmission is performed in descending order of the block priority values stored in the block.
- step S111 It is determined whether or not the recognition data transmission process is to be terminated, such as turning off the power or ending watching. When it is determined that the recognition data transmission process is continued (NO in step S111), the process proceeds to step S100, and the next captured video is acquired. If it is determined that the recognition data transmission process has ended (YES in step S111), the process ends.
- the action information of the “ fallen” action of the person A in FIG. 8 is shown as block 4 (indicated by a thick frame).
- B's “walked” action information (block 5) is transmitted.
- the emergency information can be quickly notified to the watching side by transmitting the action information before the other action information.
- the person A starts to fall, lies on the floor or the ground, and the “ fallen” action is ended when the person A does not move for a certain period of time. Therefore, it takes time to detect the falling action, and in some cases, the situation where the detection time of the “falling” of the person A is delayed after the end of the “walking” action of the person B occurs. In this case, the action information of the person B “walked” is transmitted first. Therefore, when the behavior detection at the time of abnormality is performed, it is necessary to transmit to the watching side at the earliest possible stage.
- the action information may be transmitted at the stage where the foreseeing can be foreseen.
- ID8 (FIG. 6A, fallen) is stored in the action ID of FIG. 0xFF (emergency data) is stored as the priority. Then, the value 4 as the block ID and other information are stored in the same manner as described above, and the action information “ fallen” of the person A is transmitted as the block 4 before the action information “walked” of the person B.
- ID3 (FIG. 6A, walked) is stored in the action ID of FIG. 5, and 0x80 (normal data) is stored in the block priority. Then, the value 5 as the block ID and other information are stored in the same manner as described above, and the action information of the person B “walked” is transmitted as the block 5.
- FIG. 9 is a flowchart showing the flow of recognition data transmission processing at the time of abnormality according to the present embodiment.
- Step S120 It is determined whether or not the recognition information recognized from the video data of the imaging device 10 is behavior information at the time of abnormality such as “fall down” or “no longer present” registered in advance. Since the action information at the time of abnormality often requires an emergency, the transmission order of the blocks is changed. If the action information is emergency data (Yes in step S120), the process proceeds to step S121. If the action information is not emergency data (No in step S120), the process proceeds to step S108. (Step S121) When the recognized action information is emergency, a value of 0xFF indicating emergency data is stored in the block priority field in the block. In order to show that the action information detected by this is emergency data, it becomes the block transmitted first in step S110.
- the transmission of behavior information in the event of an abnormality in FIG. 8 not only the abnormality is notified to the watching side, but also a still image is transmitted together with the behavior information in order to inform the situation by video.
- the behavior information is transmitted as the block 4, but after the transmission of the block 4, a still image obtained by photographing the “falling” situation corresponding to the block 4 is transmitted as the block 5.
- the action information (block 6) of the person B “walked” is transmitted.
- the necessary information (including still images) is transmitted to a minimum compared to the monitoring services and surveillance cameras that are currently used for video transmission. It is only necessary to reduce the amount of transmission data. Furthermore, when watching, if all the daily action information is received, a huge amount of data is obtained. When only video data is transmitted, it is necessary to view the video all day, which is very inconvenient. Therefore, in the normal behavior, behavior information representing the behavior is transmitted, and when an abnormality occurs, the corresponding video is transmitted together with the behavior information, thereby improving the convenience of watching and reducing the amount of data transfer In turn, power consumption can be reduced.
- the still image may be a thumbnail image.
- ID8 (falling) is stored in the action ID of FIG. 11 and the block priority is 0xFF (emergency Data). Then, a video (still image) obtained by photographing the “falling” situation of the person A is recorded in a format described later, and the video (still image) data is transmitted as the block ID 5.
- the ID of the block 5 that is the transmitted video data is stored in the additional information area of FIG. (Video block information) After the text information in the additional information area, the number of videos in the sixth field and the video block ID in the seventh field are stored.
- the video block ID stores a block ID of a block in which video data (still image) corresponding to the detected behavior information is stored.
- the recording area for storing one video block ID has a fixed length of 3 bytes, and a plurality of these can be stored in order. Therefore, the video block ID area has a variable length size.
- the number of stored video block IDs is stored in the video number field.
- the recording area of the video number field is 1 byte, and video block IDs from 1 to 255 can be stored.
- a value of 0 in the number-of-videos field indicates that there is no video block ID, indicating that no data is stored in the video block ID area and there is no video data corresponding to action information.
- the value 5 that is the ID of the block 5 in which the video data corresponding to the block 4 is stored is stored in the field of the block ID 1 of the video, and the value 1 that is the number of video data recorded in the number of videos is stored.
- the example in which the block 5 is transmitted as video data corresponding to one piece of behavior information is shown, but regardless of this, two or more pieces of video data are transmitted for one piece of behavior information. Also good.
- the block ID corresponding to the number of video data to be transmitted is stored in the video block ID field, and the number of data is stored in the video number field.
- the video data when two or more videos (still images or videos) taken at different angles or videos of different times are transmitted with respect to one action information, etc. Is mentioned.
- the video data when the video data is a moving image, it can be used when one video data is divided into a plurality of blocks and transmitted, or when both a still image and a moving image are transmitted.
- Block 5 Video data transmission
- FIG. 12 is a diagram for explaining the order and contents of data stored from the top of a block for transmitting video data.
- the data stored in the block is binary data in accordance with the format of FIG.
- Each information stored in the block is in the order of block ID, video data flag, corresponding ID, codec ID, data size, and video data from the top.
- Block ID The block ID of the first field is a unique ID added to each block, and is stored in order from 0 for the block to be transmitted.
- the block ID recording area has a fixed length of 3 bytes, and stores a value from 0 to 0xFFFFFF.
- “5” is stored in the block ID.
- the video data flag in the second field stores information indicating whether the block to be transmitted is video data or action information.
- the recording area of the video data flag has a fixed length of 1 byte, and a value 0 is stored in the case of a block storing video data.
- the block field storing the action information corresponding to the video data flag has the block priority. Since the block priority stores a value of 1 or more, the 3rd byte from the head of the block storing the field information is stored. By determining whether the value of the 1-byte area is 0 or any other value, it is possible to determine whether it is a block of video data or a block of behavior information.
- the priority of the video data can be acquired by referring to the block priority of the block ID indicated by the corresponding ID field described later.
- the correspondence ID of the third field stores the block ID of action information referring to the video data block.
- the recording area of the corresponding ID has a fixed length of 3 bytes, and stores values of action information block ID 0 to 0xFFFFFF referring to the video data block.
- “4” which is the ID of the block 4 of the action information referring to the block 5 storing the video data is stored.
- the codec ID of the fourth field stores codec information of the stored video data.
- the recording area of the codec ID has a fixed length of 1 byte, and stores an ID representing the type of codec shown in FIG.
- the value 1 is stored as the codec ID.
- the value 8 is stored as the codec ID.
- the ID representing the encoding method is used as the codec information.
- the format format of the container that stores the video data may be stored.
- H.M In the case of the MP4 format storing video data generated by the H.264 codec, the value 12 is stored as the codec ID.
- the codec information it is possible to obtain the codec information by interpreting the metadata of the MP4 data stored in the video data after interpreting the codec ID 12 and recognizing the MP4 format. It becomes.
- JPEG data is used as the still image of the video data corresponding to the action information of the block 4
- “3” is stored as the codec ID.
- codec ID 0 indicates that there is no codec information.
- codec IDs 14 to 255 are secured as reserved areas. When codec information is added later, codec IDs 14 to 255 are used to newly define the type of codec.
- Data size As the data size of the fifth field, the data size of the stored video data is stored. The data size recording area has a fixed length of 4 bytes, and stores size information in units of bytes. The maximum data size that can be stored is 4 GB. If the video data size is larger than 4 GB, the data may be divided into a plurality of blocks and transmitted as described above.
- Video data Video data is stored in the video data of the sixth field.
- the recording area of the video data has a variable length of 0 to 4 GB, and stores the video data from the top of the video data field.
- the value 0 is stored in the data size field, and nothing is stored in the video data field.
- the data size of the entire block is 12 bytes that is the sum of the recording areas of the data size from the block ID.
- Block 6 Behavior information transmission
- ID3 (walked) is stored in the action ID of FIG. 11, and 0x80 (normal data) is stored in the block priority.
- the value 6 is stored as the block ID and other information is stored in the same manner as described above, and the action information of the person B “walked” is transmitted as the block 6 after the blocks 4 and 5.
- the watching side It is possible to quickly notify the behavior information at the time of abnormality.
- the moving image (of the period) corresponding to the behavior may be transmitted as block data.
- the moving image corresponding to the action is stored as video data in the video data field of FIG. 12, and the codec information is stored in the codec ID (for example, MPEG having a codec ID of 8).
- the watching side analyzes the received action information and its additional information, and if the codec is a moving picture such as MPEG, the video data stored in the video data field is sent to the decoder, and the moving picture data is used using the codec indicated by the codec ID. May be played.
- the transmitted still images and videos may be transmitted in association with normal behavior such as laughing as well as behavior at the time of abnormality.
- a smiley image By transmitting a smiley image, safety confirmation can be performed together with the video, so that it is possible to improve the sense of security of the watching side.
- the video to be transmitted for each action may be set in advance according to the request from the watching side, and the video may be transmitted according to the action.
- the setting here is information such as whether or not to transmit a video in accordance with an action, and whether or not to transmit a video as a still image or a moving image.
- the block 5 storing the video data is transmitted immediately after the block 4, but the block 5 may be transmitted after the behavior information of the block 6 is transmitted.
- the block 5 may be transmitted before or after the block 4, or may be transmitted between the block 4 and another block.
- the behavior information and video data blocks may be transmitted in any order.
- the action information can refer to the corresponding block of video data by the video block ID field
- the video data can refer to the block ID of the action information referring to the video data by the corresponding ID field. Therefore, it becomes possible to refer to the corresponding blocks regardless of the transmission order of the blocks.
- FIG. 14 is a flowchart showing recognition data transmission processing at the time of abnormality according to the present embodiment.
- Step S100 to S111 and S120 to S121 are the same as described above, and thus the description thereof is omitted.
- Step S131 When detecting behavior information in an emergency, transmitting the video data is the best way to watch the situation and inform the side of the situation. For this reason, an image in which an emergency action is detected is stored in the aforementioned video data block and transmitted.
- step S132 When emergency action information is detected and video data is transmitted (Yes in step S131), the process proceeds to step S132.
- the emergency action information is transmitted and the video data is not transmitted (No in step 131)
- the process proceeds to step S110.
- Step S132 When transmitting video data in an emergency, after transmitting the block storing the action information, the block storing the corresponding video data is transmitted.
- the video data blocks are transmitted in the order of the video block ID stored in the behavior information.
- the process proceeds to step S110.
- the behavior information at the time of abnormality requiring emergency is detected, not only the behavior information but also the video data can be quickly transmitted. Even if the watcher receives a notification of action information that requires urgent action, the degree and importance of the action information are difficult to understand, but by transmitting the video, it is possible to quickly take appropriate measures. In addition to emergency video data, it is possible to give a sense of security to the watching side by transmitting an image when a normal life occurs or an image that is enjoyed.
- the behavior of a person is recognized by the method of the present embodiment and the behavior information is transmitted in units of the behavior, the amount of data can be reduced and the convenience on the side of watching can be improved.
- the behavior information is stored with higher priority than other information and is transmitted first, so that the behavior at the abnormal time can be monitored. Can be notified quickly.
- the event of an abnormality by adding the video, it is possible to quickly provide a means of confirming the safety to the watching side, reduce the amount of transmission data, and perform watching with an efficient transmission amount. it can.
- the watching side by transmitting to the watching side the person's characteristic data and action data detected from the video, the watching side does not always watch the transmission video, for example, watching a TV that has been woken up in the morning. It is possible to receive notification of information that the person is living safely such as being in a safe state, or information that an abnormality such as falling is occurring.
- the amount of transmission data can be reduced by transmitting the feature data and action data of the person detected from the video in units of action of the person.
- the transmission control unit 14 may be realized by a computer.
- the program for realizing the control function may be recorded on a computer-readable recording medium, and the program recorded on the recording medium may be read by a computer system and executed.
- the “computer system” is a computer system built in the recognition data transmission apparatus, and includes an OS and hardware such as peripheral devices.
- the “computer-readable recording medium” refers to a storage device such as a flexible medium, a magneto-optical disk, a portable medium such as a ROM or a CD-ROM, and a hard disk incorporated in a computer system. Furthermore, the “computer-readable recording medium” is a medium that dynamically holds a program for a short time, such as a communication line when transmitting a program via a network such as the Internet or a communication line such as a telephone line, In such a case, a volatile memory inside a computer system serving as a server or a client may be included and a program that holds a program for a certain period of time.
- the program may be a program for realizing a part of the functions described above, and may be a program capable of realizing the functions described above in combination with a program already recorded in a computer system.
- a part or all of the recognition data transmission apparatus in the above-described embodiment may be realized as an integrated circuit such as an LSI (Large Scale Integration).
- LSI Large Scale Integration
- Each functional block of the recognition data transmission apparatus may be individually made into a processor, or a part or all of them may be integrated into a processor.
- the method of circuit integration is not limited to LSI, and may be realized by a dedicated circuit or a general-purpose processor. Further, in the case where an integrated circuit technology that replaces LSI appears due to progress in semiconductor technology, an integrated circuit based on the technology may be used.
- Each component of the present invention can be arbitrarily selected, and an invention having a selected configuration is also included in the present invention.
- the present invention includes the following disclosure.
- (Appendix) (1) At least one imaging unit; A person detection unit that detects a feature quantity of a person from the video imaged by the imaging unit, and identifies a person in the video from the detected feature quantity; An action recognition unit for recognizing a person's action including information related to a person's movement or person from a feature amount related to the person detected by the person detection unit; A data conversion unit that divides the recognition information output by the behavior recognition unit on the time axis for each human behavior, and generates divided block data; A recognition data transmission apparatus comprising: a transmission control unit that performs control to transmit the block data output from the data conversion unit to the outside as recognition data for each block.
- a recognition data transmission apparatus comprising:
- the present invention by detecting the person detected from the video and the action thereof and transmitting the information in units of the person's action, it is possible to achieve both improvement in convenience on the side of watching and reduction of transmission data. (3) Furthermore, only the recognition information which the said action recognition part outputs is output, and it transmits outside, The recognition data transmission apparatus as described in (1) or (2) characterized by the above-mentioned.
- the transmission control unit Recognizing that the detection of the action has ended with reference to the time when the action has ended, and transmitting the action information in the order in which the detection has ended, any one of (1) to (3) The recognition data transmission device described.
- One action is transmitted from the start to the end of the action as the minimum unit of action information.
- the behavior information and the video data may be referred to by the start time and the end time of the behavior, so that the video corresponding to the behavior can be searched.
- the transmission control unit The recognition according to (4), wherein when the behavior ends, the behavior information is set in a descending order of behavior start time, and the behavior information is transmitted in descending order of priority. Data transmission equipment.
- the behavior information may be transmitted in descending order of priority.
- (6) Of the recognition information output by the action recognition unit when detecting action information that requires an emergency (including the fact that a person is missing or falls and cannot move), the action information is divided for each action of the person. Stores emergency information indicating that it is urgent in the block data, The transmission control unit The recognition data transmission device according to any one of (1) to (5), wherein the block data having the emergency information is controlled to be transmitted before other block data.
- the behavior recognition unit outputs video data captured by the imaging unit corresponding to the behavior information when the behavior information requiring the emergency is detected,
- the transmission control unit transmits the video data corresponding to the action information requiring emergency before the other block data excluding the block data storing the action information requiring emergency (6).
- the recognition data transmission device described.
- the recognition data transmission apparatus wherein the data conversion unit stores the action information requiring the emergency and information that allows the corresponding video data to be referred to each other.
- the user information output by the user information detection unit is feature information of a person (information including a person's face position, face orientation, gender, age, facial expression, and human body part) (2) To the recognition data transmission device according to any one of (8).
- the recognition information output by the behavior recognition unit is information indicating behavior (including a person standing, sitting, sleeping, falling down, or missing).
- the recognition data transmission device according to any one of (1) to (9).
- (11) At least one imaging unit; A person detection unit for detecting a feature amount of a person (part or whole of the body) from the image captured by the imaging unit, and identifying a person in the image from the detected feature amount; An action recognition unit for recognizing a person's action including information related to a person's movement or person from a feature amount related to the person detected by the person detection unit; A data conversion unit that divides the recognition information output by the behavior recognition unit on the time axis for each human behavior, and generates divided block data; A recognition data recording apparatus comprising: a recording control unit that causes the recording unit to record the block data output from the data conversion unit as recognition data.
- At least one imaging unit At least one imaging unit; A person detection unit for detecting a feature amount of a person (part or whole of the body) from the image captured by the imaging unit, and identifying a person in the image from the detected feature amount; A user information detection unit for detecting user information representing a physical feature of a person from a feature amount related to the person detected by the person detection unit; A motion detection unit that detects a user's motion (movement or gesture gesture) from the user information output by the user information detection unit and an image captured by the imaging unit; An action recognition unit that recognizes an action including a person's movement from the motion information output by the motion detection unit and the user information; A data conversion unit that divides the recognition information output by the behavior recognition unit on the time axis for each human behavior, and generates divided block data; A recognition data recording apparatus comprising
- a recognition data transmission method comprising: a transmission control step of performing control to transmit the block data output in the data conversion step to the outside as recognition data for each block.
- An action recognition step for recognizing an action including a motion of a person from the motion information output in the motion detection step and the user information;
- a recognition data transmission method comprising: a transmission control step of performing control to transmit the block data output in the data conversion step to the outside as recognition data for each block.
- the program for making a computer perform the recognition data transmission method as described in said (13) or (14).
- the computer-readable recording medium which stores the program of said (15).
- a data conversion step of dividing the recognition information output in the action recognition step on the time axis for each human action, and generating divided block data;
- a recognition data recording method comprising: a recording control step for causing the recording unit to record the block data output by the data conversion step as recognition data.
- An action recognition step for recognizing an action including a movement of a person from the motion information output in the motion detection step and the user information;
- a recognition data recording method comprising: a recording control step for causing the recording unit to record the block data output by the data conversion step as recognition data.
- the present invention can be used as a recognition data transmission device.
- DESCRIPTION OF SYMBOLS 10 ... Imaging device 101, 102-n ... Imaging part, 11 ... Image
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Alarm Systems (AREA)
- Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)
- Image Processing (AREA)
- Closed-Circuit Television Systems (AREA)
- Telephonic Communication Services (AREA)
- Image Analysis (AREA)
Abstract
Description
(第1の実施形態)
本実施の形態は、宅内における人の位置や行動を認識するセンサを用い、そのセンサが取得した情報を遠隔地にいる人に知らせる見守りや子守り、侵入者の監視、防犯、安心安全などのセキュリティへの利用、または宅内の家電機器制御に利用するための画像処理装置、撮像装置、画像処理方法およびプログラムに関するものである。例えば、行動記録装置、行動再生装置、行動検出撮像装置などが該当する。
〔見守りシステム〕
図1は、本発明の第1の実施形態に係る認識データ伝送装置を用いた見守りシステムの概念図である。
〔行動情報伝送〕
図2は、認識データ伝送装置1が検知した行動情報の伝送例を示す図である。行動を検知される人物がA、B2人の場合について図示している。図2の左から右に向けて時間の経過を表している。最初に人物Aが「入室した」行動を行い、「入室した」行動が終了した後、一定時間経過後に「立った」行動を行ったことを表している。ここで、図面上の左右方向の矢印は対応する行動に要した経過時間を表している。このように、図2においては、人物A、Bの行動について経過時間とともに、時系列に示している。
〔認識データ伝送装置〕
図3を参照して本実施の形態による認識データ伝送装置の構成例について説明する。
〔映像解析部〕
図4を用いて映像解析部11の一構成例について詳細に説明する。
〔動きベクトル検出部〕
動きベクトル検出部110は、撮像装置10から入力される複数の映像から映像内の動いた領域を検出する。1つの撮像部の映像を時間方向に1フレーム以上ためておき、現在のフレーム画素値と対応する同座標の過去のフレームの画素値の差分値が予め定めた閾値より大きい領域が、動いた領域として検出する。動きベクトル検出部110は、検出した動いた領域とその重心座標、移動量と移動方向(ベクトル)を追跡情報(動きベクトル情報)として、人物検出部111およびモーション検出部113に入力し、同時に検出情報として出力する。
〔人物検出部〕
人物検出部111は、撮像装置10から入力される複数の映像と、動きベクトル検出部110から入力される動きベクトル情報とから顔面の画像を表す領域を検出する。人物検出部111は、検出した顔面の領域における代表点(例えば、重心点)の2次元座標や、その領域の上端、下端、左端、右端の2次元座標を表す2次元顔面領域情報を生成する。人物検出部111は、複数の撮像部が撮影した複数の画像から生成した前述の2次元座標を組合せ、室内の3次元空間座標系の3次元座標に変換して、3次元の顔位置情報を生成する。ここで顔を検出できた場合に、人物が室内に入室したと判定する。
〔ユーザ情報検出部〕
次に、ユーザ情報検出部112の構成について説明する。人物検出部111で検出した人物情報に基づき、人物の特徴量として予め登録してある情報DBのデータと比較して、属性推定部1120は検出した人物情報の年齢や性別を推定する。また、表情推定部1121は人物の笑うや泣く、怒りなどの表情を推定する。人物識別部1122は、予め情報DBに登録してある人物の顔器官の特徴量と比較し、登録者のだれと一致するかの照合を行い、映像データ内にいる人物を識別する。登録者と一致した場合は、その人物の名前を、だれとも一致しない場合は、未知の人として識別結果を出力する。位置検出部1123は、人物の映像上の位置を出力する。人体部位検出部1124は、検出した人物の頭、腕、手、足、胴体といった人体の部位を検出し、その映像中の位置情報や大きさを出力する。
〔モーション検出部〕
モーション検出部113は、撮像装置10から入力される複数の映像と、人物検出部111から入力される人物情報と、動きベクトル検出部110から入力される追跡情報(動きベクトル情報)から、検出した人の動きや行動などのモーションを検出する。人物検出部111から入力される顔の3次元空間座標系の座標位置が、予め定めた閾値より高い位置にある場合に、立っていると判断する。さらに、予め情報DBに登録した人の識別情報と一緒に、身長情報を登録しておき、人物検出部111から出力される識別情報と身長情報から、立っていると判断する顔の3次元座標の範囲を設定しておけば、より高精度に立っていると判断できる。次に、同様に顔の3次元座標が床より予め定めた閾値より床側にあるときに寝ている(もしくは倒れている)と判断する。ベッドなどが設置されている場合は、その高さを前述の閾値に含めることにより、同様に寝ているかを判断する。これら、立っているもしくは寝ている以外の状態を座っていると判断する。モーション検出部113は、これら判定結果をモーション情報として出力する。
〔行動認識部〕
行動認識部114は、ユーザ情報検出部112から入力される検出情報である室内の人の有無の検出および追跡結果や、入室者が誰であるかの識別結果、在室人のモーション検出結果と、情報DB12の情報から、入室者の行動を検出し、情報DB12に記録する。検出される行動には、入室者が立っているか、座っているか、寝ているか、部屋のどこに位置し、どちらの方向を向いているか等を検出する。また、入室者のいる位置と人の向いている方向を検出することにより、例えば、テレビを見ている、調理をしている、食事をしているなどを検出することができる。
〔行動情報〕
図5を参照しながら、ブロックで伝送する行動情報の詳細な例について説明する。
(ブロックID)
1つ目のフィールドのブロックIDは、ブロック毎に付加される固有のIDであり、伝送するブロックに対して0から順番に格納される。ブロックIDの記録領域は3Byteの固定長であり、0から0xFFFFFFの値を格納する。例えば、図2のAの最初の行動である「入室した」の行動情報を格納するブロック0のブロックIDには、「0」を格納する。
(ブロック優先度)
2つ目のフィールドのブロック優先度は、検出した行動情報の優先度を格納する。優先度が高ければ高いほど、先に(早い時間に)行われた(終了した)行動であり、「倒れた」などの緊急を要する重要な行動情報であることを示す。ブロック優先度の記録領域は1Byteの固定長であり、1から0xFFの値を格納する。値1は優先度が一番低く、値が増加するにつれ、優先度が高くなることを示す。また、0xFFで緊急データ(最優先)であることを示す。
(行動ID)
3つ目のフィールドの行動IDは、検出した行動の内容を表すIDである。行動IDの記録領域は1Byteの固定長であり、図6(a)の行動IDに示すように、行動の種類に対応するIDを格納する。映像解析部11で認識した行動情報に対応する行動の種類のIDを格納する。例えば、図2のAの最初の行動である「入室した」の場合は、ID2を格納する。
(開始時間)
4つ目のフィールドの開始時間は、行動IDで示した行動を開始した時間を示す情報を格納する。開始時間の記録領域は4Byteの固定長であり、撮像部101で撮影した映像データと、映像データから認識した行動情報を伝送する場合、撮影開始からのフレーム番号を格納する。見守りを開始してからの撮影映像に対応するフレーム番号を格納することにより、認識した行動に対応する映像をすぐに検索することが可能となる。
(行動時間)
5つ目のフィールドの行動時間は、行動IDで示した行動に費やした経過時間を示す情報を格納する。行動時間の記録領域は4Byteの固定長であり、撮像部101で撮影した映像データと、映像データから認識した行動情報を伝送する場合、行動開始(開始時間フィールドが示すフレーム番号)からのフレーム数を格納する。見守りを開始してからの撮影映像に対応するフレーム数を格納することにより、認識した行動の経過時間(行動期間)に対応する映像をすぐに検索することが可能となる。
(フレームレートID)
6つ目のフィールドのフレームレートIDは、撮像部101で撮影した映像データのフレームレートを表す情報を格納する。フレームレートIDの記録領域は1Byteの固定長であり、図6(b)に示すフレームレートの種類に対応したIDを格納する。例えば、映像データが1秒間に60フレームの場合は、IDの値8を格納する。図6(b)では、0から13のフレームレートの種類を示しているが、これだけのフレームレートの種類に関わらず他のフレームレートがあってもよい。そのため、フレームレートが後で追加できるように、ID14から255までReserved領域として確保する。フレームレートの追加が発生した場合は、ID14から255に追加で定義すれば良い。
(付加情報)
7つ目のフィールドの付加情報サイズ、および8つ目のフィールドの付加情報は、映像解析部11が出力する認識情報を決定するのに用いる各検出情報を格納する。付加情報の記録領域は0から65535Byteの可変長であり、伝送する付加情報のサイズを付加情報サイズに格納する。付加情報サイズの記録領域は2Byteの固定長であり、付加情報サイズの大きさをByte単位で格納する。
(行動信頼度)
付加情報の1つ目のフィールドである行動信頼度は、映像解析部11が認識した行動に対する信頼度(認識情報の確からしさ)を格納する。行動信頼度の記録領域は2Byteの固定長であり、0から65535の値を格納する。値が大きいほど、信頼度が高いことを示す。
(顔識別ID)
付加情報の2つ目のフィールドである顔識別IDは、映像解析部11で認識した行動を行った人を識別するIDを格納する。顔識別IDの記録領域は2Byteの固定長であり、0から65535の値を格納する。値0は識別情報なし(識別者なし)を表し、値1から人の識別情報を表す。識別情報は、予め識別したい人の顔を撮影し、映像解析部11の人物識別部1122にて、撮影した人の特徴を顔情報として抽出し、識別IDに割り当てることで、識別したい人を登録する。そして、見守り開始からの撮影映像から同様に人物識別部1122にて撮影した人の特徴を顔情報として抽出し、予め登録した識別IDのどのIDと一致するかを識別し、一致する識別IDが存在した場合は、そのIDを顔識別IDに格納する。一致する識別IDが存在しない(予め登録してある人と別の人)の場合は、顔識別IDに0を格納する。
(顔識別信頼度)
付加情報の3つ目のフィールドである顔識別信頼度は、映像解析部11の人物識別部1122で識別した人の識別IDに対する信頼度(認識情報の確からしさ)を格納する。顔識別信頼度の記録領域は2Byteの固定長であり、0から65535の値を格納する。値が大きいほど、信頼度が高いことを示す。
(テキスト情報)
付加情報の4つ目のフィールドのテキスト情報サイズ、および5つ目のフィールドのテキスト情報は、伝送ブロックに関する説明や、顔識別者の名前、認識した行動に対する注意や警告に対応する通知内容等の文字情報を格納する。テキスト情報の記録領域は0から255Byteの可変長であり、伝送するテキスト情報のサイズをテキスト情報サイズに格納する。テキスト情報サイズの記録領域は1Byteの固定長であり、テキストサイズの大きさをByte単位(1文字単位)で格納する。
〔処理フロー〕
次に、本実施の形態に係る認識データ伝送処理について説明する。
(ステップS100)
映像解析部11が、撮像装置10から映像データを取得する。次に、ステップS101に進む。
(ステップS101)
動きベクトル検出部110が、撮像装置10で撮像された映像データに基づいて、撮影時間が異なる複数のフレーム映像を保持しておき、映像上で画素値が変化した領域を動いた領域として検出し、その移動量を出力する。検出した移動量や移動座標、移動ベクトルを動きベクトル情報として出力する。次のフレーム映像の動き量を検出するため、現在のフレーム映像を情報DB(メモリ)12に保存する。次に、ステップS102に進む。
(ステップS102)
人物検出部111が、撮像装置10で撮像された映像データと、ステップS101で検出した動きベクトルを使って、映像データ内の人物を検出する。映像データ内から前述の方法により顔を検出し、顔が存在した場合は、人物が存在したと判定する。また、映像中に人物がいる場合でも、必ずしも顔が撮影されるとは限らないため、算出した動きベクトルを用いて、動いた領域に人物がいると判定してもよい。さらに、顔検出情報と動きベクトルを組み合わせて人物を検出してもよい。ステップS101で出力された動きベクトルとステップS102で検出した人物情報とを照合しながら、検出した人物を追跡し、次のフレームでの人物検出精度の向上や顔検出範囲を特定することにより計算量の削減が可能となる。映像中の人物の有無、映像内の人物の位置、顔の大きさ、顔の向き、顔器官の特徴量、人物の追跡結果といった人物情報を出力し、ステップS103に進む。
(ステップS103)
ユーザ情報検出部112が、ステップS102で算出した人物情報に基づき、人物の特徴量として予め登録してある情報DBのデータと比較して、属性推定部1120は検出した人物情報の年齢や性別を推定する。また、表情推定部1121は人物の笑いや泣く、怒りなどの表情を推定する。人物識別部1122は、予め情報DBに登録してある人物の顔器官の特徴量と比較し、登録者のだれと一致するかの照合を行い、映像データ内にいる人物を識別する。登録者と一致した場合は、その人物の名前を、だれとも一致しない場合は、未知の人として識別結果を出力する。位置検出部1123は、人物の映像上の位置を出力する。人体部位検出部1124は、検出した人物の頭、腕、手、足、胴体といった人体の部位を検出し、その映像中の位置情報や大きさを出力する。ステップ103では、これら検出したユーザ情報を出力し、ステップS104に進む。
(ステップS104)
モーション検出部113が、撮影時間の異なる複数のフレーム映像から検出した人物の位置情報と人体部位情報、動きベクトル情報、人物追跡情報とから、人物の動きや身振り手振りを含むモーション情報を検出する。そして、ステップS105に進む。
(ステップS105)
行動認識部114が、撮像装置10で撮像した映像データと、検出した人物情報、ユーザ情報、モーション情報に基づいて、人の立っている、座っている、寝ている、倒れているなどの人物の行動に関する情報(行動情報)を認識し、認識情報として出力する。その後、ステップS106に進む。
(ステップS106)
ステップS105の行動認識処理から出力された認識情報を情報DB(メモリ)に保存する。これは、複数のフレーム間で認識を行う動き情報や追跡情報、モーション情報の検出を行うために使用され、現在の映像データから検出した認識情報を保存する。その後、ステップS107に進む。
(ステップS107)
ステップS106から出力される認識情報を伝送するため、人物の行動毎に認識情報を分割し、1つの行動に対する認識情報を1つのブロックデータとして生成する。1つのブロックデータは、前述の行動開始、終了時間や行動に対応する行動IDなどである。そして、ステップS108に進む。
(ステップS108)
入力された映像データで検出された行動について、本映像データ内で行動が終了したことを検出した、すなわち、同一時刻で終了した行動情報が他にあるかを判定する。同一時刻に行動が終了した行動情報がある場合(ステップS108 Yes)、ステップS109に進む。同一時刻に行動が終了した行動情報が他にない場合(ステップS108 No)、行動情報を格納するブロックのブロック優先度フィールドに0x80(通常データ)を格納し、ステップS110に進む。
(ステップS109)
同一時刻に終了した行動情報がある場合、その中で行動開始時刻が早い順に、行動情報を格納するブロックのブロック優先度フィールドに優先度の高い値を格納する。例えば、行動開始時刻が一番遅かった行動情報に対しては、ブロック優先度に0x80(通常データ)を格納する。次に行動開始時刻が早かった行動情報に対しては、ブロック優先度に0x81を格納するというように、行動開始時刻が早い行動情報の方がより高い優先度の値をブロック優先度に格納する。また、もし行動開始時刻が同じ行動情報があった場合は、予め優先度の高い人物を設定しておき、優先度の高い人物の行動には、より高い優先度の値をブロック優先度に格納するようにする。ブロック優先度の値を格納し終えたら、ステップS110に進む。
(ステップS110)
映像データから検出された行動情報をブロック単位で伝送する。伝送する順番は、ブロック内に格納されたブロック優先度の値が高い順に伝送する。全てのブロックの行動情報を伝送し終えたら、ステップS111に進む。
(ステップS111)
電源OFFや見守り終了など、認識データ伝送処理を終了させるか否かを判断する。認識データ伝送処理を引き続き継続すると判断された場合(ステップS111 NO)、ステップS100に進み、次の撮影映像を取得する。認識データ伝送処理が終了したと判断された場合(ステップS111 YES)、処理を終了する。
(第2の実施形態)
〔異常時の伝送〕
次に、図2に記載の行動情報に追加して、異常時の行動情報を伝送する場合の1例について図8を参照しながら説明する。
〔行動情報〕
次に、本実施の形態において伝送されるブロックに格納する行動情報について説明する。
〔処理フロー〕
次に、本実施の形態に係る異常時の認識データ伝送処理について説明する。
(ステップS120)
撮像装置10の映像データから認識した認識情報が予め登録してある「倒れた」や「居なくなった」などの異常時の行動情報であったか否かを判定する。異常時の行動情報は緊急を要する場合が多いため、ブロックの伝送順番を変更する。行動情報が緊急時のデータであった場合(ステップS120でYes)、ステップS121に進む。行動情報が緊急時のデータでなかった場合(ステップS120でNo)、ステップS108に進む。
(ステップS121)
認識した行動情報が緊急時である場合は、ブロック内ブロック優先度フィールドに緊急データであることを示す0xFFの値を格納する。これで検出した行動情報が緊急データであることを示すため、ステップS110で最も先に伝送されるブロックとなる。
(第3の実施形態)
〔静止画の伝送〕
次に図8の異常時の行動を認識した場合の行動情報に追加して、対応する静止画を伝送する場合の例について、図10を参照しながら説明する。
〔ブロック4:行動情報伝送〕
次に、本実施形態で伝送するブロックに格納する行動情報について説明する。
(映像ブロック情報)
付加情報領域内のテキスト情報の後に、6つ目のフィールドの映像数、および7つ目のフィールドの映像ブロックIDを格納する。映像ブロックIDは、検出した行動情報に対応する映像データ(静止画像)が格納されたブロックのブロックIDを格納する。1つの映像ブロックIDを格納する記録領域は、3Byteの固定長であり、これらを順に複数個格納することができる。したがって、映像ブロックID領域は、可変長サイズとなる。格納した映像ブロックIDの数を映像数フィールドに格納する。映像数フィールドの記録領域は、1Byteであり、1から255までの映像ブロックIDを格納することができる。映像数フィールドが値0の場合は、映像ブロックIDが存在しないことを表し、映像ブロックID領域にデータが格納されず、行動情報に対応する映像データが存在しないことを示す。
〔ブロック5:映像データ伝送〕
次に、映像データを伝送するブロックのデータ構造について、図12を参照して説明する。
(ブロックID)
1つ目フィールドのブロックIDは、ブロック毎に付加される固有のIDであり、伝送するブロックに対して0から順番に格納される。ブロックIDの記録領域は3Byteの固定長であり、0から0xFFFFFFの値を格納する。ここでは、ブロック4に対応する映像データとしてブロック5を伝送するため、ブロックIDには、「5」を格納する。
(映像データフラグ)
2つ目フィールドの映像データフラグは、伝送するブロックが映像データであるか、行動情報であるかを示す情報を格納する。映像データフラグの記録領域は1Byteの固定長であり、映像データを格納したブロックの場合は、値0を格納する。映像データフラグに対応する行動情報を格納するブロックのフィールドは、ブロック優先度になるが、ブロック優先度は1以上の値を格納するため、これらフィールド情報が格納されているブロックの先頭から3Byte目の1Byte領域の値が0かそれ以外であるかを判定することによって、映像データのブロックであるか行動情報のブロックであるかの判定をすることができる。また、映像データの優先度を取得する場合は、後述する対応IDフィールドが示すブロックIDのブロック優先度を参照することにより可能となる。
(対応ID)
3つ目のフィールドの対応IDは、映像データブロックを参照している行動情報のブロックIDを格納する。対応IDの記録領域は3Byte固定長であり、映像データブロックを参照した行動情報のブロックID0から0xFFFFFFの値を格納する。ここでは、映像データを格納したブロック5を参照している行動情報のブロック4のIDである「4」を格納する。
(コーデックID)
4つ目のフィールドのコーデックIDは、格納した映像データのコーデック情報を格納する。コーデックIDの記録領域は1Byteの固定長であり、図13に示すコーデックの種類を表すIDを格納する。例えば、伝送する映像データが静止画像の場合でBitmapデータの場合は、コーデックIDとして値1を格納する。伝送する映像データが動画の場合でMPEGの場合は、コーデックIDとして値8を格納する。ここでの説明では、コーデック情報として符号化方式を表すIDを用いて説明したが、映像データを格納するコンテナのフォーマット形式を格納してもよい。例えば、H.264のコーデックで生成された映像データを格納しているMP4フォーマットの場合は、コーデックIDとして値12を格納する。この場合、コーデック情報を取得するには、コーデックID12を解釈し、MP4フォーマットと認識した上で、映像データに格納されたMP4データのメタデータを解釈することで、コーデック情報を取得することが可能となる。ここでは、ブロック4の行動情報に対応する映像データの静止画として、例えばJPEGデータを用いるとすると、コーデックIDとして「3」を格納する。
(データサイズ)
5つ目のフィールドのデータサイズは、格納した映像データのデータサイズを格納する。データサイズの記録領域は4Byteの固定長であり、Byte単位のサイズ情報を格納する。格納できるデータサイズは、最大4GByteである。もし、映像データサイズが4GByteを超えるような大きいサイズのデータを伝送する場合は、前述のように、複数のブロックに分割して伝送すればよい。
(映像データ)
6つ目のフィールドの映像データには、映像データを格納する。映像データの記録領域は、0から4GByteの可変長であり、映像データフィールドの先頭から映像データを格納する。映像データを格納しないブロックを伝送する場合は、データサイズフィールドに値0を格納し、映像データフィールドには、何も格納しない。この場合、ブロック全体のデータサイズは、ブロックIDからデータサイズの記録領域を合計した12Byteである。
〔ブロック6:行動情報伝送〕
次に、人物Bの「歩いた」の行動終了後に、図11の行動IDに、ID3(歩いた)を格納し、ブロック優先度に0x80(通常データ)を格納する。そして、ブロックIDとして値6と、前述と同様にその他情報を格納し、人物Bの「歩いた」の行動情報をブロック6として、ブロック4および5の後に伝送する。
〔処理フロー〕
次に、本実施の形態に係る異常時の認識データ伝送処理について説明する。
(ステップS131)
緊急時の行動情報を検出した場合に、それがどのような状況かを見守り側に知らせるには、その映像データを伝送することが一番良い方法である。そのため、緊急時の行動を検出した画像を前述の映像データブロックに格納し、伝送する。緊急時の行動情報を検出し、映像データを伝送する場合(ステップS131でYes)、ステップS132に進む。緊急時の行動情報を伝送し、映像データを伝送しない場合(ステップ131でNo)、ステップS110に進む。
(ステップS132)
緊急時の映像データを伝送する場合は、その行動情報を格納したブロックを伝送した後に、対応する映像データを格納したブロックを伝送する。伝送する映像データは複数ある場合があるため、行動情報のブロックを伝送した後に、映像データブロックを、行動情報に格納した映像ブロックIDの順番に伝送する。全ての行動情報のブロックと映像データのブロックを伝送し終えたら、ステップS110に進む。
(付記)
(1)
少なくとも1つの撮像部と、
前記撮像部が撮像した映像から人物の特徴量を検出し、検出した前記特徴量から前記映像中の人物を特定する人物検出部と、
前記人物検出部が検出した人物に関する特徴量から
人物の動きや人物に関わる情報を含む人物の行動を認識する行動認識部と、
前記行動認識部が出力する認識情報を人の行動毎に時間軸上で分割し、分割したブロックデータを生成するデータ変換部と、
前記データ変換部が出力する前記ブロックデータを、認識データとしてブロック毎に外部に伝送する制御を行う伝送制御部と
を有することを特徴とする認識データ伝送装置。
(2)
少なくとも1つの撮像部と、
前記撮像部が撮像した映像から人物(の身体の一部、または全部)の特徴量を検出し、検出した前記特徴量から前記映像中の人物を特定する人物検出部と、
前記人物検出部が検出した人物に関する特徴量から人物の身体的特徴を表すユーザ情報を検出するユーザ情報検出部と、
前記ユーザ情報検出部が出力する前記ユーザ情報と、前記撮像部で撮像した映像と、からユーザのモーション(動きまたは身振り手振り)を検出するモーション検出部と、
前記モーション検出部が出力するモーション情報と、前記ユーザ情報とから、人物の動きを含む行動を認識する行動認識部と、
前記行動認識部が出力する認識情報を人の行動毎に時間軸上で分割し、分割したブロックデータを生成するデータ変換部と、
前記データ変換部が出力する前記ブロックデータを、認識データとしてブロック毎に外部に伝送する制御を行う伝送制御部と
を有することを特徴とする認識データ伝送装置。
(3)
さらに、前記行動認識部が出力する認識情報だけを出力し、外部に伝送することを特徴とする(1)または(2)に記載の認識データ伝送装置。
(4)
前記伝送制御部は、
前記行動が終了した時点を基準にしてその行動の検知が終了したと認識し、検知が終了した順番に行動情報を伝送することを特徴とする(1)から(3)までのいずれか1に記載の認識データ伝送装置。
(5)
前記伝送制御部は、
前記行動が終了した時点が同じ場合には、行動開始時刻が早い順に行動情報の優先度を高く設定し、優先度の高い順に行動情報を伝送することを特徴とする(4)に記載の認識データ伝送装置。
(6)
前記行動認識部が出力する前記認識情報のうち、(人物が居なくなっている、倒れて動けなくなっている、を含む)緊急を要する行動情報を検知したときは、前記人の行動毎に分割したブロックデータに緊急であることを示す緊急情報を格納し、
前記伝送制御部は、
前記緊急情報を有するブロックデータを、その他のブロックデータより先に伝送するように制御することを特徴とする(1)から(5)までのいずれか1に記載の認識データ伝送装置。
(7)
前記行動認識部は、前記緊急を要する行動情報を検知したときに、その行動情報に対応する前記撮像部が撮像した映像データを出力し、
前記伝送制御部は、前記緊急を要する行動情報を格納したブロックデータを除くその他のブロックデータより先に、前記緊急を要する行動情報と対応する映像データを伝送することを特徴とする(6)に記載の認識データ伝送装置。
(8)
前記データ変換部は、前記緊急を要する行動情報と、対応する映像データを互いに参照可能とする情報を格納することを特徴とする(7)に記載の認識データ伝送装置。
(9)
前記ユーザ情報検出部が出力する前記ユーザ情報は、人物の特徴情報(人物の顔の位置、顔の向き、性別、年齢、表情、人体部位を含む情報)であることを特徴とする(2)から(8)までのいずれか1に記載の認識データ伝送装置。
(10)
前記行動認識部が出力する前記認識情報は、(人物が立っている、座っている、寝ている、倒れている、居なくなっている、を含む)行動を示す情報であることを特徴とする(1)から(9)までのいずれか1に記載の認識データ伝送装置。
(11)
少なくとも1つの撮像部と、
前記撮像部が撮像した映像から人物(の身体の一部、または全部)の特徴量を検出し、検出した前記特徴量から前記映像中の人物を特定する人物検出部と、
前記人物検出部が検出した人物に関する特徴量から
人物の動きや人物に関わる情報を含む人物の行動を認識する行動認識部と、
前記行動認識部が出力する認識情報を人の行動毎に時間軸上で分割し、分割したブロックデータを生成するデータ変換部と、
前記データ変換部が出力する前記ブロックデータを、認識データとして記録部に記録させる記録制御部と
を有することを特徴とする認識データ記録装置。
(12)
少なくとも1つの撮像部と、
前記撮像部が撮像した映像から人物(の身体の一部、または全部)の特徴量を検出し、検出した前記特徴量から前記映像中の人物を特定する人物検出部と、
前記人物検出部が検出した人物に関する特徴量から人物の身体的特徴を表すユーザ情報を検出するユーザ情報検出部と、
前記ユーザ情報検出部が出力する前記ユーザ情報と、前記撮像部で撮像した映像と、からユーザのモーション(動きまたは身振り手振り)を検出するモーション検出部と、
前記モーション検出部が出力するモーション情報と、前記ユーザ情報とから、人物の動きを含む行動を認識する行動認識部と、
前記行動認識部が出力する認識情報を人の行動毎に時間軸上で分割し、分割したブロックデータを生成するデータ変換部と、
前記データ変換部が出力する前記ブロックデータを、認識データとして記録部に記録させる記録制御部と
を有することを特徴とする認識データ記録装置。
(13)
少なくとも1つの撮像部が撮像した映像から人物の特徴量を検出し、検出した前記特徴量から前記映像中の人物を特定する人物検出ステップと、
前記人物検出ステップで出力される人物に関する特徴量から人物の動きや人物に関わる情報を含む人物の行動を認識する行動認識ステップと、
前記行動認識ステップで出力される認識情報を人の行動毎に時間軸上で分割し、分割したブロックデータを生成するデータ変換ステップと、
前記データ変換ステップが出力する前記ブロックデータを、認識データとしてブロック毎に外部に伝送する制御を行う伝送制御ステップと
を有することを特徴とする認識データ伝送方法。
(14)
少なくとも1つの撮像部が撮像した映像から人物の特徴量を検出し、検出した前記特徴量から前記映像中の人物を特定する人物検出ステップと、
前記人物検出ステップにおいて検出した人物に関する特徴量から人物の身体的特徴を表すユーザ情報を検出するユーザ情報検出ステップと、
前記ユーザ情報検出ステップにおいて出力する前記ユーザ情報と、前記撮像部で撮像した映像と、からユーザのモーション(動きまたは身振り手振り)を検出するモーション検出ステップと、
前記モーション検出ステップで出力されるモーション情報と、前記ユーザ情報とから、人物の動きを含む行動を認識する行動認識ステップと、
前記行動認識ステップで出力される認識情報を人の行動毎に時間軸上で分割し、分割したブロックデータを生成するデータ変換ステップと、
前記データ変換ステップが出力する前記ブロックデータを、認識データとしてブロック毎に外部に伝送する制御を行う伝送制御ステップと
を有することを特徴とする認識データ伝送方法。
(15)
上記(13)または(14)に記載の認識データ伝送方法をコンピュータに実行させるためのプログラム。
(16)
上記(15)のプログラムを格納するコンピュータ読み取り可能な記録媒体。
(17)
少なくとも1つの撮像部が撮像した映像から人物(の身体の一部、または全部)の特徴量を検出し、検出した前記特徴量から前記映像中の人物を特定する人物検出ステップと、
前記人物検出ステップにおいて出力された人物に関する特徴量から人物の動きや人物に関わる情報を含む人物の行動を認識する行動認識ステップと、
前記行動認識ステップで出力された認識情報を人の行動毎に時間軸上で分割し、分割したブロックデータを生成するデータ変換ステップと、
前記データ変換ステップが出力する前記ブロックデータを、認識データとして記録部に記録させる記録制御ステップと
を有することを特徴とする認識データ記録方法。
(18)
少なくとも1つの撮像部が撮像した映像から人物(の身体の一部、または全部)の特徴量を検出し、検出した前記特徴量から前記映像中の人物を特定する人物検出ステップと、
前記人物検出ステップで検出された人物に関する特徴量から人物の身体的特徴を表すユーザ情報を検出するユーザ情報検出ステップと、
前記ユーザ情報検出ステップで出力された前記ユーザ情報と、前記撮像部で撮像した映像と、からユーザのモーション(動きまたは身振り手振り)を検出するモーション検出ステップと、
前記モーション検出ステップにおいて出力されたモーション情報と、前記ユーザ情報とから、人物の動きを含む行動を認識する行動認識ステップと、
前記行動認識ステップで出力された認識情報を人の行動毎に時間軸上で分割し、分割したブロックデータを生成するデータ変換ステップと、
前記データ変換ステップが出力する前記ブロックデータを、認識データとして記録部に記録させる記録制御ステップと
を有することを特徴とする認識データ記録方法。
(19)
上記(17)または(18)に記載の認識データ記録方法をコンピュータに実行させるためのプログラム。
(20)
上記(19)のプログラムを格納するコンピュータ読み取り可能な記録媒体。
Claims (5)
- 少なくとも1つの撮像部と、
前記撮像部が撮像した映像から人物の特徴量を検出し、検出した前記特徴量から前記映像中の人物を特定する人物検出部と、
前記人物検出部が検出した人物に関する特徴量から人物の動きや人物に関わる情報を含む人物の行動を認識する行動認識部と、
前記行動認識部が出力する認識情報を人の行動毎に時間軸上で分割し、分割したブロックデータを生成するデータ変換部と、
前記データ変換部が出力する前記ブロックデータを、認識データとしてブロック毎に外部に伝送する制御を行う伝送制御部と
を有することを特徴とする認識データ伝送装置。 - 前記伝送制御部は、
前記行動が終了した時点を基準にしてその行動の検知が終了したと認識し、検知が終了した順番に行動情報を伝送することを特徴とする請求項1に記載の認識データ伝送装置。 - 前記伝送制御部は、
前記行動が終了した時点が同じ場合には、行動開始時刻が早い順に行動情報の優先度を高く設定し、優先度の高い順に行動情報を伝送することを特徴とする請求項2に記載の認識データ伝送装置。 - 前記行動認識部が出力する前記認識情報のうち、緊急を要する行動情報を検知したときは、前記人の行動毎に分割したブロックデータに緊急であることを示す緊急情報を格納し、
前記伝送制御部は、
前記緊急情報を有するブロックデータを、その他のブロックデータより先に伝送するように制御することを特徴とする請求項1から3までのいずれか1項に記載の認識データ伝送装置。 - 前記行動認識部は、前記緊急を要する行動情報を検知したときに、その行動情報に対応する前記撮像部が撮像した映像データを出力し、
前記伝送制御部は、前記緊急を要する行動情報を格納したブロックデータを除くその他のブロックデータより先に、前記緊急を要する行動情報と対応する映像データを伝送することを特徴とする請求項4に記載の認識データ伝送装置。
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US15/105,452 US10699541B2 (en) | 2013-12-17 | 2014-12-08 | Recognition data transmission device |
JP2015553487A JP6411373B2 (ja) | 2013-12-17 | 2014-12-08 | 認識データ伝送装置、認識データ記録装置及び認識データ記録方法 |
CN201480065305.9A CN105794191B (zh) | 2013-12-17 | 2014-12-08 | 识别数据传输装置及方法和识别数据记录装置及方法 |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2013-260425 | 2013-12-17 | ||
JP2013260425 | 2013-12-17 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2015093330A1 true WO2015093330A1 (ja) | 2015-06-25 |
Family
ID=53402679
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2014/082371 WO2015093330A1 (ja) | 2013-12-17 | 2014-12-08 | 認識データ伝送装置 |
Country Status (4)
Country | Link |
---|---|
US (1) | US10699541B2 (ja) |
JP (1) | JP6411373B2 (ja) |
CN (1) | CN105794191B (ja) |
WO (1) | WO2015093330A1 (ja) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2017034617A (ja) * | 2015-08-06 | 2017-02-09 | 三菱電機株式会社 | 機器制御装置、機器制御システム、機器制御方法及び機器制御プログラム |
JP2018139403A (ja) * | 2016-12-21 | 2018-09-06 | アクシス アーベー | ビデオ監視システムで警告を発する方法 |
JP2018163430A (ja) * | 2017-03-24 | 2018-10-18 | セコム株式会社 | ウェアラブル端末、安否確認システム、安否確認方法およびプログラム |
JP2019179523A (ja) * | 2018-03-30 | 2019-10-17 | ダイキン工業株式会社 | 心身状態認識システム |
JP2020522828A (ja) * | 2017-04-28 | 2020-07-30 | チェリー ラボ,インコーポレイテッド | コンピュータービジョンベースの監視システムおよび方法 |
JP2022501891A (ja) * | 2018-09-20 | 2022-01-06 | ノキア テクノロジーズ オーユー | 人工知能についての装置及び方法 |
WO2022185569A1 (ja) * | 2021-03-02 | 2022-09-09 | 株式会社日立製作所 | 映像解析システム、及び映像解析方法 |
JP7524106B2 (ja) | 2021-03-02 | 2024-07-29 | 株式会社日立製作所 | 映像解析システム、及び映像解析方法 |
Families Citing this family (29)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP6669064B2 (ja) * | 2014-06-19 | 2020-03-18 | 日本電気株式会社 | 情報提示装置、情報提示システム、情報提示方法、および、プログラム |
JP6669948B2 (ja) * | 2014-09-08 | 2020-03-18 | 日本電産株式会社 | 移動体制御装置及び移動体 |
US9474042B1 (en) | 2015-09-16 | 2016-10-18 | Ivani, LLC | Detecting location within a network |
US11350238B2 (en) | 2015-09-16 | 2022-05-31 | Ivani, LLC | Systems and methods for detecting the presence of a user at a computer |
US10455357B2 (en) | 2015-09-16 | 2019-10-22 | Ivani, LLC | Detecting location within a network |
US10321270B2 (en) | 2015-09-16 | 2019-06-11 | Ivani, LLC | Reverse-beacon indoor positioning system using existing detection fields |
US10382893B1 (en) * | 2015-09-16 | 2019-08-13 | Ivani, LLC | Building system control utilizing building occupancy |
US10665284B2 (en) | 2015-09-16 | 2020-05-26 | Ivani, LLC | Detecting location within a network |
US11533584B2 (en) | 2015-09-16 | 2022-12-20 | Ivani, LLC | Blockchain systems and methods for confirming presence |
US10325641B2 (en) | 2017-08-10 | 2019-06-18 | Ivani, LLC | Detecting location within a network |
US10277836B2 (en) * | 2016-03-07 | 2019-04-30 | Ricoh Company, Ltd. | Communication terminal, image management apparatus, image processing system, method for controlling display, and computer program product |
US10380429B2 (en) | 2016-07-11 | 2019-08-13 | Google Llc | Methods and systems for person detection in a video feed |
US10957171B2 (en) | 2016-07-11 | 2021-03-23 | Google Llc | Methods and systems for providing event alerts |
JP2018173763A (ja) * | 2017-03-31 | 2018-11-08 | 積水化学工業株式会社 | 行動支援システム、行動支援方法 |
US11783010B2 (en) | 2017-05-30 | 2023-10-10 | Google Llc | Systems and methods of person recognition in video streams |
US10599950B2 (en) | 2017-05-30 | 2020-03-24 | Google Llc | Systems and methods for person recognition data management |
EP3410413B1 (en) * | 2017-06-02 | 2021-07-21 | Netatmo | Improved generation of alert events based on a detection of objects from camera images |
US11134227B2 (en) | 2017-09-20 | 2021-09-28 | Google Llc | Systems and methods of presenting appropriate actions for responding to a visitor to a smart home environment |
US10664688B2 (en) | 2017-09-20 | 2020-05-26 | Google Llc | Systems and methods of detecting and responding to a visitor to a smart home environment |
CN108082478B (zh) * | 2017-12-16 | 2021-11-16 | 深圳市翼飞鸿天无人机科技有限公司 | 一种预警型无人机监视侦查*** |
JP7104563B2 (ja) * | 2018-06-08 | 2022-07-21 | シャープ株式会社 | 動物識別装置、動物識別装置の制御方法およびプログラム |
CN108921098B (zh) * | 2018-07-03 | 2020-08-18 | 百度在线网络技术(北京)有限公司 | 人体运动分析方法、装置、设备及存储介质 |
CN110263743B (zh) * | 2019-06-26 | 2023-10-13 | 北京字节跳动网络技术有限公司 | 用于识别图像的方法和装置 |
KR102213859B1 (ko) * | 2019-08-23 | 2021-02-08 | ㈜ 한국공학기술연구원 | 이동식 감시 장치를 이용한 모니터링 시스템 |
US11893795B2 (en) | 2019-12-09 | 2024-02-06 | Google Llc | Interacting with visitors of a connected home environment |
CN113037311A (zh) * | 2021-02-24 | 2021-06-25 | 重庆工程职业技术学院 | 一种用于室内定位导航的手环 |
CN112949512B (zh) * | 2021-03-08 | 2022-07-08 | 豪威芯仑传感器(上海)有限公司 | 一种动态手势识别方法、手势交互方法及交互*** |
CN113645504B (zh) * | 2021-07-30 | 2022-08-02 | 深圳市悦道科技有限公司 | 一种tv应用软件管理***及方法 |
KR102391575B1 (ko) * | 2021-12-29 | 2022-04-28 | 주식회사 쉐카이나 | 비전카메라를 이용한 크레인 안전관리 시스템 |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006260524A (ja) * | 2005-02-16 | 2006-09-28 | Nonomura Denshi Giken:Kk | 介護通報装置 |
JP2009033564A (ja) * | 2007-07-27 | 2009-02-12 | Sanyo Electric Co Ltd | 表示装置及び表示プログラム |
JP2009199516A (ja) * | 2008-02-25 | 2009-09-03 | Sogo Keibi Hosho Co Ltd | 動作検出装置および動作検出方法 |
JP2010204758A (ja) * | 2009-02-27 | 2010-09-16 | Omron Corp | 撮影対象選択装置、撮影対象選択装置の制御方法、撮影対象選択装置制御プログラムおよび該プログラムを記録したコンピュータ読み取り可能な記録媒体 |
JP2012004720A (ja) * | 2010-06-15 | 2012-01-05 | Hitachi Ltd | 監視装置、及び監視システム |
Family Cites Families (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6049281A (en) * | 1998-09-29 | 2000-04-11 | Osterweil; Josef | Method and apparatus for monitoring movements of an individual |
JP4736171B2 (ja) * | 2000-10-20 | 2011-07-27 | 日本ビクター株式会社 | 監視用信号の通信方法、及び監視用信号通信装置 |
US6678413B1 (en) * | 2000-11-24 | 2004-01-13 | Yiqing Liang | System and method for object identification and behavior characterization using video analysis |
JP2005318445A (ja) | 2004-04-30 | 2005-11-10 | Funai Electric Co Ltd | 遠隔見守りシステムの宅内装置 |
US8704668B1 (en) * | 2005-04-20 | 2014-04-22 | Trevor Darrell | System for monitoring and alerting based on animal behavior in designated environments |
JP2007243342A (ja) * | 2006-03-06 | 2007-09-20 | Yokogawa Electric Corp | 画像監視装置及び画像監視システム |
US7642035B2 (en) * | 2006-04-13 | 2010-01-05 | Industrial Technology Research Institute | Sensitized photochemical switching for cholesteric liquid crystal displays |
JP4389971B2 (ja) * | 2007-06-19 | 2009-12-24 | ミツミ電機株式会社 | 指紋画像形成装置及びこれに用いる指移動量推定方法 |
JP2009009413A (ja) * | 2007-06-28 | 2009-01-15 | Sanyo Electric Co Ltd | 動作検知装置及び動作検知プログラム、並びに動作基本モデル生成装置及び動作基本モデル生成プログラム |
US7557759B2 (en) * | 2007-07-02 | 2009-07-07 | Cheng Uei Precision Industry Co., Ltd. | Integrated multi-band antenna |
US20090033564A1 (en) * | 2007-08-02 | 2009-02-05 | Nigel Power, Llc | Deployable Antennas for Wireless Power |
JP4716139B2 (ja) * | 2008-05-14 | 2011-07-06 | アイシン精機株式会社 | 周辺監視装置 |
JP4636135B2 (ja) | 2008-08-04 | 2011-02-23 | ソニー株式会社 | 画像処理装置、撮像装置、画像処理方法およびプログラム |
US9886761B2 (en) | 2012-06-08 | 2018-02-06 | Sony Corporation | Information processing to display existing position of object on map |
-
2014
- 2014-12-08 WO PCT/JP2014/082371 patent/WO2015093330A1/ja active Application Filing
- 2014-12-08 US US15/105,452 patent/US10699541B2/en active Active
- 2014-12-08 JP JP2015553487A patent/JP6411373B2/ja active Active
- 2014-12-08 CN CN201480065305.9A patent/CN105794191B/zh active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006260524A (ja) * | 2005-02-16 | 2006-09-28 | Nonomura Denshi Giken:Kk | 介護通報装置 |
JP2009033564A (ja) * | 2007-07-27 | 2009-02-12 | Sanyo Electric Co Ltd | 表示装置及び表示プログラム |
JP2009199516A (ja) * | 2008-02-25 | 2009-09-03 | Sogo Keibi Hosho Co Ltd | 動作検出装置および動作検出方法 |
JP2010204758A (ja) * | 2009-02-27 | 2010-09-16 | Omron Corp | 撮影対象選択装置、撮影対象選択装置の制御方法、撮影対象選択装置制御プログラムおよび該プログラムを記録したコンピュータ読み取り可能な記録媒体 |
JP2012004720A (ja) * | 2010-06-15 | 2012-01-05 | Hitachi Ltd | 監視装置、及び監視システム |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2017034617A (ja) * | 2015-08-06 | 2017-02-09 | 三菱電機株式会社 | 機器制御装置、機器制御システム、機器制御方法及び機器制御プログラム |
JP2018139403A (ja) * | 2016-12-21 | 2018-09-06 | アクシス アーベー | ビデオ監視システムで警告を発する方法 |
JP7229662B2 (ja) | 2016-12-21 | 2023-02-28 | アクシス アーベー | ビデオ監視システムで警告を発する方法 |
JP2018163430A (ja) * | 2017-03-24 | 2018-10-18 | セコム株式会社 | ウェアラブル端末、安否確認システム、安否確認方法およびプログラム |
JP2020522828A (ja) * | 2017-04-28 | 2020-07-30 | チェリー ラボ,インコーポレイテッド | コンピュータービジョンベースの監視システムおよび方法 |
JP2019179523A (ja) * | 2018-03-30 | 2019-10-17 | ダイキン工業株式会社 | 心身状態認識システム |
JP2022501891A (ja) * | 2018-09-20 | 2022-01-06 | ノキア テクノロジーズ オーユー | 人工知能についての装置及び方法 |
US11442985B2 (en) | 2018-09-20 | 2022-09-13 | Nokia Technologies Oy | Apparatus and a method for artificial intelligence |
WO2022185569A1 (ja) * | 2021-03-02 | 2022-09-09 | 株式会社日立製作所 | 映像解析システム、及び映像解析方法 |
JP7524106B2 (ja) | 2021-03-02 | 2024-07-29 | 株式会社日立製作所 | 映像解析システム、及び映像解析方法 |
Also Published As
Publication number | Publication date |
---|---|
JPWO2015093330A1 (ja) | 2017-03-16 |
CN105794191B (zh) | 2019-04-09 |
CN105794191A (zh) | 2016-07-20 |
US20160335861A1 (en) | 2016-11-17 |
JP6411373B2 (ja) | 2018-10-24 |
US10699541B2 (en) | 2020-06-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP6411373B2 (ja) | 認識データ伝送装置、認識データ記録装置及び認識データ記録方法 | |
US11120559B2 (en) | Computer vision based monitoring system and method | |
US11735018B2 (en) | Security system with face recognition | |
US10083599B2 (en) | Remote user interface and display for events for a monitored location | |
US9396400B1 (en) | Computer-vision based security system using a depth camera | |
JP2018120644A (ja) | 識別装置、識別方法およびプログラム | |
JP5390322B2 (ja) | 画像処理装置、及び画像処理方法 | |
JP5674406B2 (ja) | 自律移動体を用いた監視システム、監視装置、自律移動体、監視方法、及び監視プログラム | |
JP6339445B2 (ja) | 人物識別装置 | |
CN106162073A (zh) | 通信设备 | |
JP2007272488A (ja) | 画像処理装置、監視カメラ及び画像監視システム | |
JP4622301B2 (ja) | 監視システム、および監視カメラ | |
WO2013069565A1 (ja) | 撮影記録装置 | |
JP6941950B2 (ja) | 画像提供システム、画像提供方法、および画像提供プログラム | |
EP3616095A1 (en) | Computer vision based monitoring system and method | |
US20110176025A1 (en) | Video information processing apparatus, video information processing method, and computer-readable storage medium | |
WO2006123331A2 (en) | Monitoring method and device | |
KR102544147B1 (ko) | 영상분석 기반 1인 낙상 감지 시스템 및 방법 | |
WO2020008726A1 (ja) | 対象物体検出プログラム、および対象物体検出装置 | |
JP2016021716A (ja) | 追尾装置及びその制御方法 | |
WO2022249635A1 (ja) | 行動検知システムおよび行動検知プログラム | |
JP7465644B2 (ja) | 監視システム、及び監視方法 | |
KR20220077639A (ko) | 실내형 촬영디바이스 시스템 및 실내형 촬영디바이스 시스템의 동작 방법 | |
WO2019056245A1 (zh) | 家庭安防方法和终端 | |
JP2017045423A (ja) | 映像処理装置、映像処理方法、映像処理システム、およびプログラム |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 14872172 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2015553487 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 15105452 Country of ref document: US |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 14872172 Country of ref document: EP Kind code of ref document: A1 |