CN106650662A - Target object occlusion detection method and target object occlusion detection device - Google Patents
Target object occlusion detection method and target object occlusion detection device Download PDFInfo
- Publication number
- CN106650662A CN106650662A CN201611191940.9A CN201611191940A CN106650662A CN 106650662 A CN106650662 A CN 106650662A CN 201611191940 A CN201611191940 A CN 201611191940A CN 106650662 A CN106650662 A CN 106650662A
- Authority
- CN
- China
- Prior art keywords
- destination object
- occlusion detection
- thermodynamic chart
- detection method
- neural networks
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/53—Recognition of crowd images, e.g. recognition of crowd congestion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/161—Detection; Localisation; Normalisation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Human Computer Interaction (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a target object occlusion detection method and a target object occlusion detection device. The target object occlusion detection method comprises steps: input images are received; a well-trained neural network is used to detect a target image in the input images and output a thermodynamic chart of the target object; and based on the thermodynamic chart, whether occlusion exists in the target object is detected. According to the target object occlusion detection method and the target object occlusion detection device provided by the embodiment of the invention, the target object in the input images is converted to the thermodynamic chart based on the well-trained neural network, whether the occlusion exists in the target object is detected through the thermodynamic chart, the target object recognition difficulty can be effectively reduced, and the recognition precision and the stability can be enhanced.
Description
Technical field
The present invention relates to technical field of image processing, relates more specifically to a kind of destination object occlusion detection method and dress
Put.
Background technology
For the identification of certain object in image, in the presence of image is one serious dry to blocking for the object
Mouth mask, glasses, bang, cap, decorative pattern when disturbing factor, such as recognition of face on face etc. can cause to block to face.It is another
Aspect, in some applications, in the recognition of face such as in monitor video, it may appear that disturbed intentionally by blocking, resist identification
Situation.These will all have a strong impact on the precision and stability of identification.
Due to the mode blocked and type it is varied, limit is difficult to, if blocking species to each trains one
Grader, can cause workload huge, and seriously be constrained by data volume.Accordingly, it would be desirable to a kind of can be in detection image
Object is with the presence or absence of the technology or system blocked.
The content of the invention
The present invention is proposed to solve the above problems.According to an aspect of the present invention, there is provided a kind of destination object hides
Gear detection method, the destination object occlusion detection method includes：Receives input image；Using the neutral net detection for training
Destination object in the input picture simultaneously exports the thermodynamic chart of the destination object；And institute is detected based on the thermodynamic chart
Destination object is stated with the presence or absence of blocking.
In one embodiment of the invention, the neutral net for training is full convolutional neural networks.
In one embodiment of the invention, the full convolutional neural networks are based on the convolutional neural networks that will be trained
Full articulamentum replaces with convolutional layer and generates.
In one embodiment of the invention, the convolutional layer is 1 × 1 convolutional layer.
In one embodiment of the invention, the full convolutional neural networks also include up-sampling layer.
In one embodiment of the invention, the number of the up-sampling layer depends on the expectation of the thermodynamic chart of the output
Resolution ratio.
In one embodiment of the invention, the destination object occlusion detection method also includes：Exporting the heating power
After figure, the destination object in the thermodynamic chart is labeled；And it is described to be with the presence or absence of the detection blocked to destination object
Based on the thermodynamic chart after marking described in.
In one embodiment of the invention, described whether there is based on the thermodynamic chart detection destination object is blocked
The step of include：Whether the shape for detecting the destination object in the thermodynamic chart meets the object of the classification of the destination object
Shape；And when the shape of the destination object in the thermodynamic chart meets the shape of the object of the classification of the destination object,
Determine that the destination object is not present to block, otherwise, it is determined that the destination object presence is blocked.
In one embodiment of the invention, described whether there is based on the thermodynamic chart detection destination object is blocked
The step of include：Whether the region for detecting the destination object in the thermodynamic chart is connected region；And the heat ought be detected
When the region of the destination object in trying hard to is connected region, determines that the destination object is not present and block, otherwise, it is determined that it is described
Destination object presence is blocked.
In one embodiment of the invention, the destination object occlusion detection method also includes：When detecting the heat
When destination object presence in trying hard to is blocked, occlusion area is determined.
According to a further aspect of the invention, there is provided a kind of destination object occlusion detection device, the destination object blocks inspection
Surveying device includes：Receiver module, for receives input image；Thermodynamic chart output module, for using the neutral net for training
Detect the destination object in the input picture and export the thermodynamic chart of the destination object；And occlusion detection module, it is used for
Detect that the destination object whether there is based on the thermodynamic chart to block.
In one embodiment of the invention, the neutral net for training is full convolutional neural networks.
In one embodiment of the invention, the full convolutional neural networks are based on the convolutional neural networks that will be trained
Full articulamentum replaces with convolutional layer and generates.
In one embodiment of the invention, the convolutional layer is 1 × 1 convolutional layer.
In one embodiment of the invention, the full convolutional neural networks also include up-sampling layer.
In one embodiment of the invention, the number of the up-sampling layer depends on the expectation of the thermodynamic chart of the output
Resolution ratio.
In one embodiment of the invention, the destination object occlusion detection device also includes：Automatic marking module, uses
Destination object in the thermodynamic chart is labeled；And the occlusion detection module is additionally operable to be based on after marking described in
Thermodynamic chart detect the destination object with the presence or absence of blocking.
In one embodiment of the invention, the occlusion detection module is additionally operable to：Detect the target in the thermodynamic chart
Whether the shape of object meets the shape of the object of the classification of the destination object；And when the destination object in the thermodynamic chart
Shape meet the destination object classification object shape when, determine that the destination object is not present and block, conversely, then
Determine that the destination object is present to block.
In one embodiment of the invention, the occlusion detection module is additionally operable to：Detect the target in the thermodynamic chart
Whether the region of object is connected region；And when the region for detecting the destination object in the thermodynamic chart is connected region
When, determine that the destination object is not present and block, otherwise, it is determined that the destination object presence is blocked.
In one embodiment of the invention, the occlusion detection module is additionally operable to：In the thermodynamic chart is detected
When destination object presence is blocked, occlusion area is determined.
Destination object occlusion detection method and device according to embodiments of the present invention will be defeated based on the neutral net for training
Enter the destination object in image and be converted to thermodynamic chart, and can effectively be dropped with the presence or absence of blocking by thermodynamic chart detected target object
The low identification difficulty to destination object, lifts accuracy of identification and stability.
Description of the drawings
The embodiment of the present invention is described in more detail by combining accompanying drawing, above-mentioned and other purposes of the present invention,
Feature and advantage will be apparent from.Accompanying drawing is used for providing further understanding the embodiment of the present invention, and constitutes explanation
A part for book, is used to explain the present invention together with the embodiment of the present invention, is not construed as limiting the invention.In the accompanying drawings,
Identical reference number typically represents same parts or step.
Fig. 1 illustrates the exemplary electron for realizing destination object occlusion detection method and apparatus according to embodiments of the present invention
The schematic block diagram of equipment；
Fig. 2 illustrates the indicative flowchart of destination object occlusion detection method according to embodiments of the present invention；
Fig. 3 A and Fig. 3 B are shown respectively showing for convolutional neural networks according to embodiments of the present invention and full convolutional neural networks
Meaning property structure chart；
Fig. 4 A and Fig. 4 B are shown respectively the example of the example of input picture and the corresponding thermodynamic chart of output；
Fig. 5 illustrates the schematic block diagram of destination object occlusion detection device according to embodiments of the present invention；And
Fig. 6 illustrates the schematic block diagram of destination object sheltering detection system according to embodiments of the present invention.
Specific embodiment
In order that the object, technical solutions and advantages of the present invention become apparent from, root is described in detail below with reference to accompanying drawings
According to the example embodiment of the present invention.Obviously, described embodiment is only a part of embodiment of the present invention, rather than this
Bright whole embodiments, it should be appreciated that the present invention is not limited by example embodiment described herein.Described in the present invention
The embodiment of the present invention, those skilled in the art's all other embodiment resulting in the case where creative work is not paid
All should fall under the scope of the present invention.
First, with reference to Fig. 1 describing the destination object occlusion detection method and apparatus for realizing the embodiment of the present invention
Exemplary electronic device 100.
As shown in figure 1, electronic equipment 100 includes one or more processors 102, one or more storage devices 104, defeated
Enter device 106, output device 108 and imageing sensor 110, these components are by bus system 112 and/or other forms
Bindiny mechanism's (not shown) interconnection.It should be noted that the component and structure of the electronic equipment 100 shown in Fig. 1 are exemplary, and
Nonrestrictive, as needed, the electronic equipment can also have other assemblies and structure.
The processor 102 can be CPU (CPU) or perform with data-handling capacity and/or instruction
The processing unit of the other forms of ability, and it is desired to perform to control other components in the electronic equipment 100
Function.
The storage device 104 can include one or more computer programs, and the computer program can
With including various forms of computer-readable recording mediums, such as volatile memory and/or nonvolatile memory.It is described easy
The property lost memory can for example include random access memory (RAM) and/or cache memory (cache) etc..It is described non-
Volatile memory can for example include read-only storage (ROM), hard disk, flash memory etc..In the computer-readable recording medium
On can store one or more computer program instructions, processor 102 can run described program instruction, to realize hereafter institute
The client functionality (realized by processor) in the embodiment of the present invention stated and/or other desired functions.In the meter
Various application programs and various data can also be stored in calculation machine readable storage medium storing program for executing, such as application program use and/or
Various data for producing etc..
The input unit 106 can be device of the user for input instruction, and can include keyboard, mouse, wheat
One or more in gram wind and touch-screen etc..
The output device 108 can export various information (such as image or sound) to outside (such as user), and
One or more in display, loudspeaker etc. can be included.
Described image sensor 110 can shoot the desired image of user (such as photo, video etc.), and will be captured
Image be stored in the storage device 104 so that other components are used.Image collecting device 110 can be camera.Should
Work as understanding, image collecting device 110 is only example, and electronic equipment 100 can not include image collecting device 110.In this feelings
Under condition, it is possible to use other image acquisition device images to be recognized, and the images to be recognized transmission electron of collection is set
Standby 100.
Exemplarily, it is electric for realizing the example of destination object occlusion detection method and apparatus according to embodiments of the present invention
Sub- equipment may be implemented as smart mobile phone, panel computer etc..
Below, destination object occlusion detection method 200 according to embodiments of the present invention will be described with reference to Fig. 2.
In step S210, receives input image.
In one embodiment, the input picture for being received can be the image of the destination object for including to be identified.Target
Object can be the object (such as face, animal, various objects) of any one classification or plurality of classes.In to input picture
Destination object be identified before, first carry out destination object occlusion detection method according to embodiments of the present invention, with determine treat
Then the destination object of identification is identified to it again, it is possible to decrease identification difficulty with the presence or absence of blocking, and improves accuracy of identification.
In one example, the input picture for being received can be the image of Real-time Collection.In other examples, received
Input picture can also be image from any source.Herein, the input picture for being received can be video data, it is also possible to
For image data.
In step S220, detect the destination object in the input picture using the neutral net for training and export institute
State the thermodynamic chart of destination object.
In one embodiment, can be generated in step based on the convolutional neural networks as general grader for training
The neutral net of the output thermodynamic chart that rapid S220 is utilized.
In this embodiment it is possible to the classification for being based on destination object to be identified determines the target pair that carry out occlusion detection
The classification of elephant, the then identification data training first with category object is got to a destination object that can recognize the category
Grader, then the neutral net of output thermodynamic chart is generated based on the grader.
For example, the convolutional neural networks of a destination object that can recognize the category can first be trained.Fig. 3 A show
The schematic diagram of convolutional neural networks 300A according to embodiments of the present invention.As shown in Figure 3A, can recognize certain classification or
The convolutional neural networks 300A of the destination object of certain multiple classification can include convolutional layer, pond layer and full articulamentum.Input figure
As classification results can be obtained after input convolutional neural networks 300A, as shown in Figure 3A.
Then, the convolutional neural networks for being trained based on this, can be by the convolutional neural networks (for example as shown in Figure 3A
Convolutional neural networks 300A) full articulamentum replace with convolutional layer next life and help convolutional neural networks, the full convolutional neural networks
Can be used as the neutral net of the output thermodynamic chart utilized in step S220 after training.As Fig. 3 B are illustrated according to the present invention
The schematic diagram of the full convolutional neural networks 300B of embodiment.As shown in Figure 3 B, full convolutional neural networks 300B can be wrapped
Include convolutional layer and pond layer.Input picture is input into after full convolutional neural networks 300B and can obtain thermodynamic chart, as shown in Figure 3 B.
In one example, the convolutional layer for replacing the full articulamentum of convolutional neural networks 300A can be 1 × 1 convolutional layer.
Using 1 × 1 convolutional layer can in the case where desired function is realized relatively reduced operand, arithmetic speed can be improved.At it
In his example, the convolutional layer for replacing the full articulamentum of convolutional neural networks 300A can also be the convolution of any other suitable yardsticks
Layer.
With continued reference to Fig. 3 B, full convolutional neural networks 300B can also include up-sampling layer, and up-sampling layer can improve defeated
The resolution ratio of the thermodynamic chart for going out, is blocked with preferably whether there is based on thermodynamic chart detected target object.In one example, on
The number of sample level can depend on the expectation resolution ratio of the thermodynamic chart of output.In other examples, it is also possible to consider other because
The plain comprehensive number that up-sampling layer is set.
In the examples described above, the step of training convolutional neural networks 300A is properly termed as the pre-training stage, in the stage instruction
White silk is capable of identify that the general category device of the classification of destination object；The step of training full convolutional neural networks 300B is properly termed as fine setting
In the stage, full convolutional neural networks 300B is generated based on convolutional neural networks 300A in the stage, and make this complete by training
Convolutional neural networks 300B becomes the neutral net of the output thermodynamic chart utilized in step S220.Full convolutional neural networks 300B
Can be regarded as from convolutional neural networks 300A extracting the new god that embedded mapping is followed by upper thermodynamic chart output layer and constitutes
Jing networks.
Above-mentioned example has been illustratively described the training of the neutral net of the output thermodynamic chart utilized in step S220
Journey, although it will appreciated by the skilled person that the convolutional Neural used when showing training in Fig. 3 A and Fig. 3 B
The schematic diagram of resulting full convolutional neural networks after network and training, but it is only exemplary, and they may be used also
Think other any appropriate structures.Additionally, the neutral net of the output thermodynamic chart utilized in step S220 also can be instructed directly
Practice and generate, without being generated later by first training general grader.
Based on the neutral net (such as full convolutional neural networks 300B as shown in Figure 3 B) for training, can be to input figure
As in destination object detected and exported corresponding thermodynamic chart, as illustrated in figures 4 a and 4b.Fig. 4 A show input picture
Example, it is assumed that destination object to be detected is face, then the neutral net for training through above-mentioned is (for example as shown in Figure 3 B
Full convolutional neural networks 300B), exportable corresponding thermodynamic chart as shown in Figure 4 B, shows heating power figure in thermodynamic chart
The destination object (i.e. face) of formula, there is as can be seen from this figure the face (such as the face thermodynamic chart of the leftmost side in Fig. 4 B) for blocking
It is significantly different with unscreened face.
Below referring back to Fig. 2 continuing on destination object occlusion detection method 200 according to embodiments of the present invention step
Suddenly.
In step S230, detect that the destination object whether there is based on the thermodynamic chart and block.
In one embodiment, whether there is the step of blocking based on thermodynamic chart detected target object can include：Detection
Whether the shape of the destination object in thermodynamic chart meets the shape of the object of the classification of the destination object；And when in thermodynamic chart
When the shape of destination object meets the shape of the object of the classification of the destination object, determine that destination object is not present and block, conversely,
Then determine that destination object is present to block.In this embodiment, " shape " is appreciated that to be face (that is, the border being made up of border
The region of composition) rather than border itself.
For example, when occlusion detection is carried out to face, if having longer bang on face or (such as scheming with sunglasses
The face of the leftmost side in 4A) etc., then " black hole " is will appear from above the human face region on thermodynamic chart (such as such as the people of Fig. 4 B leftmost sides
Shown in face) or other forms occlusion area, by detecting its shape, can be determined whether that presence is blocked.
In another embodiment, whether the region that can detect the destination object in thermodynamic chart is connected region, works as target
When object presence is blocked, it may no longer be complete connected region, but the thing that is blocked therefrom " is blocked ", so as in thermodynamic chart
In destination object region in there is black region (as shown in the black region in the human face region of the leftmost side in Fig. 4 B), because
Whether this region that may be based on the destination object in thermodynamic chart is connected region judging destination object with the presence or absence of blocking.
In other examples, it is also possible to which whether thermodynamic chart detected target object is based on by any other suitable mode
Presence is blocked.
Further, in one embodiment, first the destination object in thermodynamic chart can be carried out after output thermodynamic chart
Mark (frame to destination object for example as shown in Figure 4 B is selected), is then based on the detection of the thermodynamic chart Jing after mark and blocks.This
Sample does will can keep apart at a distance of nearer destination object, also can be by because showing less mesh farther out and in the picture away from camera lens
Mark object will also recognize that ground frame is elected, it is to avoid omit or error occurs in occlusion detection, realize more accurately occlusion detection, improve inspection
Survey the reliability of result.In one example, the mark to destination object can be realized using neutral net.Show at another
In example, it would however also be possible to employ suitable algorithm realizes the mark to destination object.In other examples, any other can also be adopted
Suitable method realizes the mark to destination object.
In yet another embodiment, when the destination object in detecting thermodynamic chart exists and blocks, may further determine that
Occlusion area.For example, occlusion area can be split, to determine blocking position to destination object, so as to for follow-up
Identification to destination object provides more reliable foundation.In the scene of real-time application, or destination object to be identified
(such as people to be identified) provides prompting as reference, informs that it is readjusted position to be identified or removes shelter, carries
The efficiency of height identification and the degree of accuracy.
Based on above description, destination object occlusion detection method according to embodiments of the present invention is based on the nerve for training
Destination object in input picture is converted to thermodynamic chart by network, and by thermodynamic chart detected target object with the presence or absence of blocking,
The identification difficulty to destination object can be effectively reduced, accuracy of identification and stability is lifted.
Exemplarily, destination object occlusion detection method according to embodiments of the present invention can be with memory and process
Realize in the unit or system of device.
Additionally, destination object occlusion detection method processing speed according to embodiments of the present invention is fast, model small volume, can be with
Easily it is deployed on the mobile devices such as smart mobile phone, panel computer, personal computer.Alternatively, according to embodiments of the present invention
Destination object occlusion detection method can also be deployed in server end (or high in the clouds).Alternatively, it is according to embodiments of the present invention
Destination object occlusion detection method is deployed at server end (or high in the clouds) and personal terminal in which can also be distributed.
The destination object occlusion detection device of another aspect of the present invention offer is described with reference to Fig. 5.Fig. 5 shows basis
The schematic block diagram of the destination object occlusion detection device 500 of the embodiment of the present invention.
As shown in figure 5, destination object occlusion detection device 500 according to embodiments of the present invention includes receiver module 510, heat
Try hard to output module 520 and occlusion detection module 530.The modules can be performed respectively above in conjunction with Fig. 2 descriptions
Each step/function of destination object occlusion detection method.Below only to each module of destination object occlusion detection device 500
Major function is described, and omits the detail content having been described above.
Receiver module 510 is used for receives input image.Thermodynamic chart output module 520 is used for using the neutral net for training
Detect the destination object in the input picture and export the thermodynamic chart of the destination object.Occlusion detection module 530 is used for base
Detect that the destination object whether there is in the thermodynamic chart to block.Receiver module 510, thermodynamic chart output module 520 and screening
The journey stored in the Running storage device 104 of processor 102 kept off in the electronic equipment that detection module 530 can be as shown in Figure 1
Sequence instructs to realize.
In one embodiment, the input picture that receiver module 510 is received can be to include destination object to be identified
Image.Destination object can be the object (such as face, animal, various objects) of any one classification or plurality of classes.
Before being identified to the destination object in input picture, first occlusion detection is carried out to it, then it is identified again, can be dropped
Low identification difficulty, improves accuracy of identification.
In one example, the input picture that receiver module 510 is received can be the image of Real-time Collection.Show at other
In example, the input picture that receiver module 510 is received can also be the image from any source.Herein, receiver module 510 is connect
The input picture of receipts can be video data, or image data.
In one embodiment, the neutral net that thermodynamic chart output module 520 is utilized is full convolutional neural networks (as schemed
Full convolutional neural networks 300B shown in 3B).Exemplarily, the full convolutional neural networks can be based on the convolution god that will be trained
The full articulamentum of Jing networks (convolutional neural networks 300A as shown in Figure 3A) replaces with convolutional layer and generates.Thermodynamic chart exports mould
The training of the neutral net that block 520 is utilized is referred to the process described by above-mentioned combination Fig. 3 A and Fig. 3 B, for sake of simplicity, this
Place repeats no more.
In one example, the neutral net that thermodynamic chart output module 520 is utilized can also include up-sampling layer, above adopt
Sample layer can improve the resolution ratio of the thermodynamic chart of output, preferably to detect target based on thermodynamic chart by occlusion detection module 530
Object is with the presence or absence of blocking.In one example, the expectation for up-sampling the thermodynamic chart that the number of layer can depend on output is differentiated
Rate.In other examples, up-sample the number of layer it is also contemplated that other factors and comprehensive arrange.
In one embodiment, occlusion detection module 530 whether there is the step blocked based on thermodynamic chart detected target object
Suddenly can include：Whether the shape of the destination object in detection thermodynamic chart meets the shape of the object of the classification of the destination object；
And when the shape of the destination object in thermodynamic chart meets the shape of the object of the classification of the destination object, determine destination object
Do not exist and block, otherwise, it is determined that destination object presence is blocked.In this embodiment, " shape " is appreciated that to be by border institute
The face (that is, the region that border is constituted) of composition rather than border itself.
In another embodiment, occlusion detection module 530 can also be used for detecting the region of the destination object in thermodynamic chart
Whether be connected region, when destination object exist block when, it may no longer be complete connected region, but be blocked thing from
In " block ", so as to occur black region in the region of the destination object in thermodynamic chart (such as the face area of the leftmost side in Fig. 4 B
Shown in black region in domain), therefore whether the region that occlusion detection module 530 can also be used for based on destination object is connected region
Domain is judging destination object with the presence or absence of blocking.
In one embodiment, destination object occlusion detection device 500 can also include automatic marking module (not in Fig. 5
In illustrate), be labeled (such as such as Fig. 4 B for the destination object in the thermodynamic chart that exported to thermodynamic chart output module 520
Shown in the frame of destination object is selected).Also, occlusion detection module 530 can be also used for based on automatic marking mould described in Jing
Thermodynamic chart detected target object after block mark whether there is blocks.Based on the mark of automatic marking module, can by a distance of compared with
Near destination object is kept apart, also can be by because showing that less destination object will also recognize that ground frame farther out and in the picture away from camera lens
Elect, it is to avoid omit or error occurs in occlusion detection, realize more accurately occlusion detection, improve the reliability of testing result.
In yet another embodiment, occlusion detection module 530 can be also used for the destination object in thermodynamic chart is detected
When presence is blocked, occlusion area is further determined that.For example, occlusion detection module 530 can split occlusion area, with true
The fixed blocking position to destination object, so as to provide more reliable foundation for the follow-up identification to destination object.In real time should
In scene, occlusion area determined by occlusion detection module 530 can also be that destination object to be identified is (such as to be identified
People) provide prompting as reference, inform that it is readjusted position to be identified or removes shelter, improve the efficiency of identification
And the degree of accuracy.
Based on above description, destination object occlusion detection device according to embodiments of the present invention is based on the nerve for training
Destination object in input picture is converted to thermodynamic chart by network, and by thermodynamic chart detected target object with the presence or absence of blocking,
The identification difficulty to destination object can be effectively reduced, accuracy of identification and stability is lifted.
Fig. 6 shows the schematic block diagram of destination object sheltering detection system 600 according to embodiments of the present invention.Target pair
As sheltering detection system 600 includes storage device 610 and processor 620.
Wherein, during storage device 610 is stored for realizing destination object occlusion detection method according to embodiments of the present invention
Corresponding steps program code.Processor 620 is used for the program code stored in Running storage device 610, to perform basis
The corresponding steps of the destination object occlusion detection method of the embodiment of the present invention, and for realizing mesh according to embodiments of the present invention
Corresponding module in mark object occlusion detection device.Additionally, destination object sheltering detection system 600 can also include IMAQ
Device (not shown in FIG. 6), it can be used for gathering input picture.Certainly, image collecting device is not required, can be direct
Receive the input of the input picture from other sources.
In one embodiment, when described program code is run by processor 620 so that destination object occlusion detection system
System 600 performs following steps：Receives input image；The target pair in the input picture is detected using the neutral net for training
As and export the thermodynamic chart of the destination object；And detect the destination object with the presence or absence of blocking based on the thermodynamic chart.
In one embodiment, the neutral net for training is full convolutional neural networks.
In one embodiment, the full convolutional neural networks are based on the full articulamentum of the convolutional neural networks that will be trained
Replace with convolutional layer and generate.
In one embodiment, the convolutional layer is 1 × 1 convolutional layer.
In one embodiment, the full convolutional neural networks also include up-sampling layer.
In one embodiment, the number of the up-sampling layer depends on the expectation resolution ratio of the thermodynamic chart of the output.
In one embodiment, destination object occlusion detection is also caused when described program code is run by processor 620
System 600 performs following steps：After the thermodynamic chart is exported, the destination object in the thermodynamic chart is labeled；And
When described program code is run by processor 620 so that destination object sheltering detection system 600 perform it is described to target pair
As being based on the thermodynamic chart after marking described in the presence or absence of the detection blocked.
In one embodiment, when described program code is run by processor 620 so that destination object occlusion detection system
What system 600 was performed detects that the destination object whether there is the step of blocking and include based on the thermodynamic chart：Detect the heating power
Whether the shape of the destination object in figure meets the shape of the object of the classification of the destination object；And when in the thermodynamic chart
Destination object shape meet the destination object classification object shape when, determine that the destination object does not have screening
Gear, otherwise, it is determined that the destination object presence is blocked.
In one embodiment, when described program code is run by processor 620 so that destination object occlusion detection system
What system 600 was performed detects that the destination object whether there is the step of blocking and include based on the thermodynamic chart：Detect the heating power
Whether the region of the destination object in figure is connected region；And when the region for detecting the destination object in the thermodynamic chart is
During connected region, determine that the destination object is not present and block, otherwise, it is determined that the destination object presence is blocked.
In one embodiment, destination object occlusion detection is also caused when described program code is run by processor 620
System 600 performs following steps：When the destination object in detecting the thermodynamic chart is present to be blocked, occlusion area is determined.
Additionally, according to embodiments of the present invention, additionally providing a kind of storage medium, program is stored on said storage
Instruction, the destination object when described program is instructed and run by computer or processor for performing the embodiment of the present invention blocks inspection
The corresponding steps of survey method, and for realizing destination object occlusion detection device according to embodiments of the present invention in respective mode
Block.The storage medium for example can include the storage card of smart phone, the memory unit of panel computer, personal computer it is hard
Disk, read-only storage (ROM), Erasable Programmable Read Only Memory EPROM (EPROM), portable compact disc read-only storage (CD-
ROM), any combination of USB storage or above-mentioned storage medium.The computer-readable recording medium can be one or
Any combination of multiple computer-readable recording mediums, such as one computer-readable recording medium includes receives input image
Computer-readable program code, another computer-readable recording medium includes detected target object and exports the meter of thermodynamic chart
The readable program code of calculation machine, the computer-readable program generation that another computer-readable recording medium is blocked comprising detection
Code.
In one embodiment, the computer program instructions can be realized according to of the invention real when being run by computer
Each functional module of the destination object occlusion detection device of example is applied, and/or can be performed according to embodiments of the present invention
Destination object occlusion detection method.
In one embodiment, the computer program instructions make computer or place by computer or processor when running
Reason device performs following steps：Receives input image；The target pair in the input picture is detected using the neutral net for training
As and export the thermodynamic chart of the destination object；And detect the destination object with the presence or absence of blocking based on the thermodynamic chart.
In one embodiment, the neutral net for training is full convolutional neural networks.
In one embodiment, the full convolutional neural networks are based on the full articulamentum of the convolutional neural networks that will be trained
Replace with convolutional layer and generate.
In one embodiment, the convolutional layer is 1 × 1 convolutional layer.
In one embodiment, the full convolutional neural networks also include up-sampling layer.
In one embodiment, the number of the up-sampling layer depends on the expectation resolution ratio of the thermodynamic chart of the output.
In one embodiment, the computer program instructions make computer or place by computer or processor when running
Reason device performs following steps：After the thermodynamic chart is exported, the destination object in the thermodynamic chart is labeled；And it is described
Computer program instructions make computer or the described of computing device be to destination object when being run by computer or processor
The detection that no presence is blocked is based on the thermodynamic chart after marking described in.
In one embodiment, the computer program instructions make computer or place by computer or processor when running
What reason device was performed detects that the destination object whether there is the step of blocking and include based on the thermodynamic chart：Detect the thermodynamic chart
In destination object shape whether meet the destination object classification object shape；And when in the thermodynamic chart
When the shape of destination object meets the shape of the object of the classification of the destination object, determine that the destination object does not have screening
Gear, otherwise, it is determined that the destination object presence is blocked.
In one embodiment, the computer program instructions make computer or place by computer or processor when running
What reason device was performed detects that the destination object whether there is the step of blocking and include based on the thermodynamic chart：Detect the thermodynamic chart
In the region of destination object whether be connected region；And when the region for detecting the destination object in the thermodynamic chart is company
During logical region, determine that the destination object is not present and block, otherwise, it is determined that the destination object presence is blocked.
In one embodiment, the computer program instructions make computer or place by computer or processor when running
Reason device performs following steps：When the destination object in detecting the thermodynamic chart is present to be blocked, occlusion area is determined.
Each module in destination object occlusion detection device according to embodiments of the present invention can be by according to of the invention real
Apply the computer program instructions that the processor operation of the electronic equipment of the destination object occlusion detection of example is stored in memory
Realize, or the meter that can be stored in the computer-readable recording medium of computer program according to embodiments of the present invention
The instruction of calculation machine is realized when being run by computer.
Destination object occlusion detection method according to embodiments of the present invention, device, system and storage medium are based on training
Whether the destination object in input picture is converted to thermodynamic chart by good neutral net, and deposited by thermodynamic chart detected target object
Blocking, can effectively reduce the identification difficulty to destination object, lifting accuracy of identification and stability.
Although the example embodiment by reference to Description of Drawings here, it should be understood that above-mentioned example embodiment is merely exemplary
, and be not intended to limit the scope of the invention to this.Those of ordinary skill in the art can wherein carry out various changes
And modification, it is made without departing from the scope of the present invention and spirit.All such changes and modifications are intended to be included in claims
Within required the scope of the present invention.
Those of ordinary skill in the art are it is to be appreciated that the list of each example with reference to the embodiments described herein description
Unit and algorithm steps, being capable of being implemented in combination in electronic hardware or computer software and electronic hardware.These functions are actually
Performed with hardware or software mode, depending on the application-specific and design constraint of technical scheme.Professional and technical personnel
Each specific application can be used different methods to realize described function, but this realization it is not considered that exceeding
The scope of the present invention.
In several embodiments provided herein, it should be understood that disclosed apparatus and method, it can be passed through
Its mode is realized.For example, apparatus embodiments described above are only schematic, for example, the division of the unit, and only
Only a kind of division of logic function, can there is other dividing mode when actually realizing, such as multiple units or component can be tied
Close or be desirably integrated into another equipment, or some features can be ignored, or do not perform.
In specification mentioned herein, a large amount of details are illustrated.It is to be appreciated, however, that the enforcement of the present invention
Example can be put into practice in the case of without these details.In some instances, known method, structure is not been shown in detail
And technology, so as not to obscure the understanding of this description.
Similarly, it will be appreciated that in order to simplify the present invention and help understand one or more in each inventive aspect, exist
To the present invention exemplary embodiment description in, the present invention each feature be grouped together into sometimes single embodiment, figure,
Or in descriptions thereof.However, the method for the present invention should be construed to reflect following intention：It is i.e. required for protection
The more features of feature that application claims ratio is expressly recited in each claim.More precisely, such as corresponding power
As sharp claim reflects, its inventive point is can be with the spy of all features less than certain disclosed single embodiment
Levy to solve corresponding technical problem.Therefore, it then follows it is concrete that thus claims of specific embodiment are expressly incorporated in this
Separate embodiments of the embodiment, wherein each claim as the present invention itself.
It will be understood to those skilled in the art that in addition to mutually exclusive between feature, any combinations pair can be adopted
All features and so disclosed any method disclosed in this specification (including adjoint claim, summary and accompanying drawing)
Or all processes or unit of equipment are combined.Unless expressly stated otherwise, this specification (will including adjoint right
Ask, make a summary and accompanying drawing) disclosed in each feature can, equivalent identical by offer or similar purpose alternative features replacing.
Although additionally, it will be appreciated by those of skill in the art that some embodiments described herein include other embodiments
In included some features rather than further feature, but the combination of the feature of different embodiments means in of the invention
Within the scope of and form different embodiments.For example, in detail in the claims, embodiment required for protection one of arbitrarily
Can in any combination mode using.
The present invention all parts embodiment can be realized with hardware, or with one or more processor operation
Software module realize, or with combinations thereof realization.It will be understood by those of skill in the art that can use in practice
Microprocessor or digital signal processor (DSP) to realize article analytical equipment according to embodiments of the present invention in some moulds
The some or all functions of block.The present invention is also implemented as the part for performing method as described herein or complete
The program of device (for example, computer program and computer program) in portion.Such program for realizing the present invention can be stored
On a computer-readable medium, or can have one or more signal form.Such signal can be from internet
Download on website and obtain, or provide on carrier signal, or provide in any other form.
It should be noted that above-described embodiment the present invention will be described rather than limits the invention, and ability
Field technique personnel can design without departing from the scope of the appended claims alternative embodiment.In the claims,
Any reference symbol between bracket should not be configured to limitations on claims.Word "comprising" is not excluded the presence of not
Element listed in the claims or step.Word "a" or "an" before element does not exclude the presence of multiple such
Element.The present invention can come real by means of the hardware for including some different elements and by means of properly programmed computer
It is existing.If in the unit claim for listing equipment for drying, several in these devices can be by same hardware branch
To embody.The use of word first, second, and third does not indicate that any order.These words can be explained and be run after fame
Claim.
The above, the only specific embodiment of the present invention or the explanation to specific embodiment, the protection of the present invention
Scope is not limited thereto, any those familiar with the art the invention discloses technical scope in, can be easily
Expect change or replacement, all should be included within the scope of the present invention.Protection scope of the present invention should be with claim
Protection domain is defined.
Claims (20)
1. a kind of destination object occlusion detection method, it is characterised in that the destination object occlusion detection method includes：
Receives input image；
Destination object in the input picture is detected using the neutral net that trains and export the heating power of the destination object
Figure；And
Detect that the destination object whether there is based on the thermodynamic chart to block.
2. destination object occlusion detection method according to claim 1, it is characterised in that the neutral net for training
For full convolutional neural networks.
3. destination object occlusion detection method according to claim 2, it is characterised in that the full convolutional neural networks base
Generate in the full articulamentum of the convolutional neural networks for training is replaced with into convolutional layer.
4. destination object occlusion detection method according to claim 3, it is characterised in that the convolutional layer is 1 × 1 convolution
Layer.
5. destination object occlusion detection method according to claim 2, it is characterised in that the full convolutional neural networks are also
Including up-sampling layer.
6. destination object occlusion detection method according to claim 5, it is characterised in that the number of the up-sampling layer takes
Certainly in the output thermodynamic chart expectation resolution ratio.
7. destination object occlusion detection method according to claim 1, it is characterised in that the destination object occlusion detection
Method also includes：
After the thermodynamic chart is exported, the destination object in the thermodynamic chart is labeled；And
Described is based on the thermodynamic chart after marking described in the presence or absence of the detection blocked to destination object.
8. the destination object occlusion detection method according to any one of claim 1-7, it is characterised in that described to be based on
The thermodynamic chart detects that the destination object whether there is the step of blocking and include：
Whether the shape for detecting the destination object in the thermodynamic chart meets the shape of the object of the classification of the destination object；With
And
When the shape of the destination object in the thermodynamic chart meets the shape of the object of the classification of the destination object, institute is determined
State destination object and do not exist and block, otherwise, it is determined that the destination object exists and blocks.
9. the destination object occlusion detection method according to any one of claim 1-7, it is characterised in that described to be based on
The thermodynamic chart detects that the destination object whether there is the step of blocking and include：
Whether the region for detecting the destination object in the thermodynamic chart is connected region；And
When the region of the destination object in detecting the thermodynamic chart is connected region, determine that the destination object does not have screening
Gear, otherwise, it is determined that the destination object presence is blocked.
10. the destination object occlusion detection method according to any one of claim 1-7, it is characterised in that the mesh
Mark object occlusion detection method also includes：
When the destination object in detecting the thermodynamic chart is present to be blocked, occlusion area is determined.
11. a kind of destination object occlusion detection devices, it is characterised in that the destination object occlusion detection device includes：
Receiver module, for receives input image；
Thermodynamic chart output module, for using the destination object in the neutral net detection input picture for training and defeated
Go out the thermodynamic chart of the destination object；And
Occlusion detection module, blocks for detecting that the destination object whether there is based on the thermodynamic chart.
12. destination object occlusion detection devices according to claim 11, it is characterised in that the nerve net for training
Network is full convolutional neural networks.
13. destination object occlusion detection devices according to claim 12, it is characterised in that the full convolutional neural networks
Generated based on the full articulamentum of the convolutional neural networks for training is replaced with into convolutional layer.
14. destination object occlusion detection devices according to claim 13, it is characterised in that the convolutional layer is volume 1 × 1
Lamination.
15. destination object occlusion detection devices according to claim 12, it is characterised in that the full convolutional neural networks
Also include up-sampling layer.
16. destination object occlusion detection devices according to claim 15, it is characterised in that the number of the up-sampling layer
Depending on the expectation resolution ratio of the thermodynamic chart of the output.
17. destination object occlusion detection devices according to claim 11, it is characterised in that the destination object blocks inspection
Surveying device also includes：
Automatic marking module, for being labeled to the destination object in the thermodynamic chart；And
The occlusion detection module is additionally operable to detect the destination object with the presence or absence of screening based on the thermodynamic chart after marking described in
Gear.
The 18. destination object occlusion detection devices according to any one of claim 11-17, it is characterised in that described
Occlusion detection module is additionally operable to：
Whether the shape for detecting the destination object in the thermodynamic chart meets the shape of the object of the classification of the destination object；With
And
When the shape of the destination object in the thermodynamic chart meets the shape of the object of the classification of the destination object, institute is determined
State destination object and do not exist and block, otherwise, it is determined that the destination object exists and blocks.
The 19. destination object occlusion detection devices according to any one of claim 11-17, it is characterised in that described
Occlusion detection module is additionally operable to：
Whether the region for detecting the destination object in the thermodynamic chart is connected region；And
When the region of the destination object in detecting the thermodynamic chart is connected region, determine that the destination object does not have screening
Gear, otherwise, it is determined that the destination object presence is blocked.
The 20. destination object occlusion detection devices according to any one of claim 11-17, it is characterised in that described
Occlusion detection module is additionally operable to：
When the destination object in detecting the thermodynamic chart is present to be blocked, occlusion area is determined.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611191940.9A CN106650662B (en) | 2016-12-21 | 2016-12-21 | Target object shielding detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611191940.9A CN106650662B (en) | 2016-12-21 | 2016-12-21 | Target object shielding detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106650662A true CN106650662A (en) | 2017-05-10 |
| CN106650662B CN106650662B (en) | 2021-03-23 |
Family
ID=58833491
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611191940.9A Active CN106650662B (en) | 2016-12-21 | 2016-12-21 | Target object shielding detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106650662B (en) |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107330900A (en) * | 2017-06-22 | 2017-11-07 | 成都品果科技有限公司 | A kind of automatic portrait dividing method |
| CN108875540A (en) * | 2018-03-12 | 2018-11-23 | 北京旷视科技有限公司 | Image processing method, device and system and storage medium |
| CN108875537A (en) * | 2018-02-28 | 2018-11-23 | 北京旷视科技有限公司 | Method for checking object, device and system and storage medium |
| CN108932456A (en) * | 2017-05-23 | 2018-12-04 | 北京旷视科技有限公司 | Face identification method, device and system and storage medium |
| CN109063695A (en) * | 2018-09-18 | 2018-12-21 | 图普科技（广州）有限公司 | A kind of face critical point detection method, apparatus and its computer storage medium |
| CN109146924A (en) * | 2018-07-18 | 2019-01-04 | 北京飞搜科技有限公司 | A kind of method for tracking target and device based on thermodynamic chart |
| CN109389599A (en) * | 2018-10-25 | 2019-02-26 | 北京阿丘机器人科技有限公司 | A kind of defect inspection method and device based on deep learning |
| CN110287760A (en) * | 2019-03-28 | 2019-09-27 | 电子科技大学 | A kind of human face five-sense-organ point occlusion detection method based on deep learning |
| CN110428394A (en) * | 2019-06-14 | 2019-11-08 | 北京迈格威科技有限公司 | Method, apparatus and computer storage medium for target mobile detection |
| CN110765815A (en) * | 2018-07-26 | 2020-02-07 | 北京京东尚科信息技术有限公司 | Method and device for detecting shielding of display rack |
| CN111046858A (en) * | 2020-03-18 | 2020-04-21 | 成都大熊猫繁育研究基地 | Image-based animal species fine classification method, system and medium |
| CN111695495A (en) * | 2020-06-10 | 2020-09-22 | 杭州萤石软件有限公司 | Face recognition method, electronic device and storage medium |
| CN112288676A (en) * | 2019-07-12 | 2021-01-29 | 富士通株式会社 | Method and device for detecting product defects |
| CN112380981A (en) * | 2020-11-12 | 2021-02-19 | 平安科技（深圳）有限公司 | Face key point detection method and device, storage medium and electronic equipment |
| WO2021088729A1 (en) * | 2019-11-04 | 2021-05-14 | 京东数科海益信息科技有限公司 | Image processing method and apparatus, electronic device and storage medium |
| WO2021098261A1 (en) * | 2019-11-22 | 2021-05-27 | 苏州浪潮智能科技有限公司 | Target detection method and apparatus |
| CN113468975A (en) * | 2021-06-09 | 2021-10-01 | 浙江大华技术股份有限公司 | Fighting behavior detection method and device |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103942535A (en) * | 2014-03-28 | 2014-07-23 | 广东威创视讯科技股份有限公司 | Multi-target tracking method and device |
| CN104978582A (en) * | 2015-05-15 | 2015-10-14 | 苏州大学 | Contour chord angle feature based identification method for blocked target |
| CN105631427A (en) * | 2015-12-29 | 2016-06-01 | 北京旷视科技有限公司 | Suspicious personnel detection method and system |
| US9424494B1 (en) * | 2016-01-28 | 2016-08-23 | International Business Machines Corporation | Pure convolutional neural network localization |
| CN106204587A (en) * | 2016-05-27 | 2016-12-07 | 孔德兴 | Multiple organ dividing method based on degree of depth convolutional neural networks and region-competitive model |
-
2016
- 2016-12-21 CN CN201611191940.9A patent/CN106650662B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103942535A (en) * | 2014-03-28 | 2014-07-23 | 广东威创视讯科技股份有限公司 | Multi-target tracking method and device |
| CN104978582A (en) * | 2015-05-15 | 2015-10-14 | 苏州大学 | Contour chord angle feature based identification method for blocked target |
| CN105631427A (en) * | 2015-12-29 | 2016-06-01 | 北京旷视科技有限公司 | Suspicious personnel detection method and system |
| US9424494B1 (en) * | 2016-01-28 | 2016-08-23 | International Business Machines Corporation | Pure convolutional neural network localization |
| CN106204587A (en) * | 2016-05-27 | 2016-12-07 | 孔德兴 | Multiple organ dividing method based on degree of depth convolutional neural networks and region-competitive model |
Non-Patent Citations (1)
| Title |
|---|
| ***: "车辆遮挡检测算法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (25)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108932456A (en) * | 2017-05-23 | 2018-12-04 | 北京旷视科技有限公司 | Face identification method, device and system and storage medium |
| CN107330900A (en) * | 2017-06-22 | 2017-11-07 | 成都品果科技有限公司 | A kind of automatic portrait dividing method |
| CN108875537A (en) * | 2018-02-28 | 2018-11-23 | 北京旷视科技有限公司 | Method for checking object, device and system and storage medium |
| CN108875540A (en) * | 2018-03-12 | 2018-11-23 | 北京旷视科技有限公司 | Image processing method, device and system and storage medium |
| CN108875540B (en) * | 2018-03-12 | 2021-11-05 | 北京旷视科技有限公司 | Image processing method, device and system and storage medium |
| CN109146924B (en) * | 2018-07-18 | 2020-09-08 | 苏州飞搜科技有限公司 | Target tracking method and device based on thermodynamic diagram |
| CN109146924A (en) * | 2018-07-18 | 2019-01-04 | 北京飞搜科技有限公司 | A kind of method for tracking target and device based on thermodynamic chart |
| CN110765815B (en) * | 2018-07-26 | 2024-05-24 | 北京京东尚科信息技术有限公司 | Display rack shielding detection method and device |
| CN110765815A (en) * | 2018-07-26 | 2020-02-07 | 北京京东尚科信息技术有限公司 | Method and device for detecting shielding of display rack |
| CN109063695A (en) * | 2018-09-18 | 2018-12-21 | 图普科技（广州）有限公司 | A kind of face critical point detection method, apparatus and its computer storage medium |
| CN109389599A (en) * | 2018-10-25 | 2019-02-26 | 北京阿丘机器人科技有限公司 | A kind of defect inspection method and device based on deep learning |
| CN110287760A (en) * | 2019-03-28 | 2019-09-27 | 电子科技大学 | A kind of human face five-sense-organ point occlusion detection method based on deep learning |
| CN110428394A (en) * | 2019-06-14 | 2019-11-08 | 北京迈格威科技有限公司 | Method, apparatus and computer storage medium for target mobile detection |
| CN110428394B (en) * | 2019-06-14 | 2022-04-26 | 北京迈格威科技有限公司 | Method, apparatus and computer storage medium for target movement detection |
| CN112288676A (en) * | 2019-07-12 | 2021-01-29 | 富士通株式会社 | Method and device for detecting product defects |
| WO2021088729A1 (en) * | 2019-11-04 | 2021-05-14 | 京东数科海益信息科技有限公司 | Image processing method and apparatus, electronic device and storage medium |
| WO2021098261A1 (en) * | 2019-11-22 | 2021-05-27 | 苏州浪潮智能科技有限公司 | Target detection method and apparatus |
| CN111046858B (en) * | 2020-03-18 | 2020-09-08 | 成都大熊猫繁育研究基地 | Image-based animal species fine classification method, system and medium |
| CN111046858A (en) * | 2020-03-18 | 2020-04-21 | 成都大熊猫繁育研究基地 | Image-based animal species fine classification method, system and medium |
| CN111695495A (en) * | 2020-06-10 | 2020-09-22 | 杭州萤石软件有限公司 | Face recognition method, electronic device and storage medium |
| CN111695495B (en) * | 2020-06-10 | 2023-11-14 | 杭州萤石软件有限公司 | Face recognition method, electronic equipment and storage medium |
| CN112380981A (en) * | 2020-11-12 | 2021-02-19 | 平安科技（深圳）有限公司 | Face key point detection method and device, storage medium and electronic equipment |
| CN112380981B (en) * | 2020-11-12 | 2024-06-28 | 平安科技（深圳）有限公司 | Face key point detection method and device, storage medium and electronic equipment |
| CN113468975A (en) * | 2021-06-09 | 2021-10-01 | 浙江大华技术股份有限公司 | Fighting behavior detection method and device |
| CN113468975B (en) * | 2021-06-09 | 2024-07-12 | 浙江大华技术股份有限公司 | Method and device for detecting racking behavior |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106650662B (en) | 2021-03-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106650662A (en) | Target object occlusion detection method and target object occlusion detection device | |
| CN106529485A (en) | Method and apparatus for obtaining training data | |
| CN106934376B (en) | A kind of image-recognizing method, device and mobile terminal | |
| CN106203305A (en) | Human face in-vivo detection method and device | |
| CN105938552A (en) | Face recognition method capable of realizing base image automatic update and face recognition device | |
| CN106254782A (en) | Image processing method and device and camera | |
| CN105631439A (en) | Human face image collection method and device | |
| CN107977665A (en) | The recognition methods of key message and computing device in a kind of invoice | |
| CN108875722A (en) | Character recognition and identification model training method, device and system and storage medium | |
| CN106845352B (en) | Pedestrian detection method and device | |
| CN107844794A (en) | Image-recognizing method and device | |
| CN108875731A (en) | Target identification method, device, system and storage medium | |
| US11100357B2 (en) | Real-time micro air-quality indexing | |
| CN110046617A (en) | A kind of digital electric meter reading self-adaptive identification method based on deep learning | |
| CN111291761B (en) | Method and device for recognizing text | |
| CN111163265A (en) | Image processing method, image processing device, mobile terminal and computer storage medium | |
| CN106131405A (en) | Image processing method and device and for its image capturing system | |
| CN108875492A (en) | Face datection and crucial independent positioning method, device, system and storage medium | |
| CN106203425A (en) | Character identifying method and device | |
| CN109242796A (en) | Character image processing method, device, electronic equipment and computer storage medium | |
| CN110310224A (en) | Light efficiency rendering method and device | |
| CN104902143A (en) | Resolution-ratio-based image de-noising method and device | |
| CN108875509A (en) | Biopsy method, device and system and storage medium | |
| CN110031697A (en) | Test method, device, system and the computer-readable medium of target identification equipment | |
| CN109977832A (en) | A kind of image processing method, device and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 100190 Beijing, Haidian District Academy of Sciences, South Road, No. 2, block A, No. 313 Applicant after: MEGVII INC. Applicant after: Beijing maigewei Technology Co., Ltd. Address before: 100190 Beijing, Haidian District Academy of Sciences, South Road, No. 2, block A, No. 313 Applicant before: MEGVII INC. Applicant before: Beijing aperture Science and Technology Ltd. |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |