CN114359761A - Climbing unbuckled safety belt recognition system based on portable cloth ball control - Google Patents
Climbing unbuckled safety belt recognition system based on portable cloth ball control Download PDFInfo
- Publication number
- CN114359761A CN114359761A CN202111615156.7A CN202111615156A CN114359761A CN 114359761 A CN114359761 A CN 114359761A CN 202111615156 A CN202111615156 A CN 202111615156A CN 114359761 A CN114359761 A CN 114359761A
- Authority
- CN
- China
- Prior art keywords
- safety belt
- climbing
- personnel
- person
- belt
- 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.)
- Pending
Links
- 230000009194 climbing Effects 0.000 title claims abstract description 51
- 239000004744 fabric Substances 0.000 title claims abstract description 24
- 238000001514 detection method Methods 0.000 claims abstract description 62
- 230000001174 ascending effect Effects 0.000 claims abstract description 27
- 230000002708 enhancing effect Effects 0.000 claims abstract description 8
- 238000000034 method Methods 0.000 claims abstract description 8
- 230000011218 segmentation Effects 0.000 claims description 18
- 230000006399 behavior Effects 0.000 claims description 16
- 230000001133 acceleration Effects 0.000 claims description 12
- 238000013441 quality evaluation Methods 0.000 claims description 7
- 239000003086 colorant Substances 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 238000013528 artificial neural network Methods 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 claims description 4
- 230000006835 compression Effects 0.000 claims description 4
- 238000007906 compression Methods 0.000 claims description 4
- 241001503987 Clematis vitalba Species 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 7
- 238000001914 filtration Methods 0.000 abstract description 7
- 230000000007 visual effect Effects 0.000 description 10
- 238000005516 engineering process Methods 0.000 description 8
- 238000013135 deep learning Methods 0.000 description 3
- 230000003203 everyday effect Effects 0.000 description 3
- 238000012937 correction Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 206010000117 Abnormal behaviour Diseases 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of computer vision, in particular to a portable cloth control ball-based identification system for a safety belt which is not tied during ascending, which comprises the following specific operation processes of S1, panoramic snapshot; s2, positioning the personnel and the climbing carrier target; s3, judging the position relation between the carrier and the personnel; s4, controlling the cloth control ball to obtain a local map of the person ascending a height; s5, segmenting the personnel instance; s6, enhancing the safety belt image; s7, evaluating the image quality; s8, safety belt wearing identification, the system realizes intelligent snapshot of an actual operation scene by controlling the movement and zoom control of a camera, can improve the identification effect on the climbing behavior of an operator and safety belt wearing, solves the problem of low accuracy of a target detection algorithm under a long distance, and increases an image quality filtering module before a safety belt detection algorithm to filter low-quality ball distribution video pictures and improve the detection effect of the safety belt detection algorithm.
Description
Technical Field
The invention relates to the technical field of computer vision, in particular to a portable cloth control ball-based identification system for a safety belt which is not fastened during ascending.
Background
Guangdong power grid company proposes to create a visual field intelligent safe business card in the whole process, the current power grid company has many operating points every day and large difficulty in field safety control, although the current operation site visual supervision and management system is popularized, the field safety management and control are carried out only by means of manpower staring at the disk, the efficiency is low, the violation of the field operation or the dangerous behaviors cannot be reminded and alarmed in real time, at present, only 4000 plus 6000 electric power operation points of the Guangdong power grid are needed every day, 4-6 million personnel of the operation field are needed every day, the operation points are multiple and wide, the difficulty of the field safety management and control is high, in the face of the severe situation, the field safety management and control are carried out only by means of manpower safety supervision, the supervision efficiency is low, the manpower cost is high, the problem of supervision blind areas easily occurs, the field situation is difficult to be comprehensively and timely mastered, and the safety risk is difficult to be strictly controlled.
The climbing without fastening the safety belt means that: the personnel who carry out high altitude construction do not wear or wear the safety belt by mistake, according to the regulation in the electric power safety work regulations of the south China electric network Limited liability company, the behavior in the operation belongs to the condition of violation punishment, so the AI image recognition technology is urgently needed to be relied on, the operation personnel can be timely reminded and supervised in the actual operation, and the violation condition of ascending an unmanned staircase is avoided.
The most similar prior art to the present invention:
1. target detection
The task of object detection is to find out interested objects in an image or a video and detect the positions and the sizes of the interested objects, and the method is one of the core problems in the field of machine vision. Since the time of deep learning, the development of object detection mainly focuses on two directions: the two main differences are that the two algorithms need to generate a proposal (a preselected frame possibly containing an object to be detected) first, and then perform fine-grained object detection, while the one-stage algorithm can directly extract features in a network to predict object classification and position, but these algorithms cannot solve the problem that a target object is very small and cannot be detected due to the fact that a camera is far away from the object.
The prior art has the following disadvantages:
1. the existing safety belt detection algorithm can cause the size of an object in a picture to be very small when the target object is far away from a camera, so that the target detection algorithm is invalid.
2. The existing safety belt detection algorithm lacks a filtering module for low-quality and unrecognizable pictures in practical application. The detection algorithm has many situations of false detection and missing detection of the safety belt in practical application.
3. The existing safety belt detection algorithm has the defects that the detection accuracy rate is reduced under the condition that the shape and the color of the safety belt are possibly similar to the colors of outer clothing and backpacks of people, and the overall color of the overall personnel picture is consistent, and an image enhancement module is not used for compensating and correcting the safety belt area image.
4. Most of the existing image enhancement algorithms enhance the whole ball distribution and control video image, so that the image quality enhancement algorithms occupy large memory and have long single-frame image processing time.
Therefore, it is very important to design a novel portable ball-distribution-control-based identification system for a climbing unbelted safety belt to overcome the technical defects.
Disclosure of Invention
The invention aims to provide a climbing unbelted belt recognition system based on a portable cloth control ball, so as to solve the problems in the background technology.
In order to achieve the purpose, the invention provides the following technical scheme:
a portable cloth control ball-based identification system for a safety belt not tied during ascending is disclosed, which comprises the following specific operation procedures:
s1, carrying out panoramic snapshot;
s2, positioning the personnel and the climbing carrier target;
s3, judging the position relation between the carrier and the personnel;
s4, controlling the cloth control ball to obtain a local map of the person ascending a height;
s5, segmenting the personnel instance;
s6, enhancing the safety belt image;
s7, evaluating the image quality;
and S8, identifying the wearing of the safety belt.
As a preferable scheme of the present invention, the panoramic capturing in S1 specifically includes:
the system can control the shooting angle and the focal length of the camera through an ONVIF protocol, ensures that an operation site is completely presented in a video picture as far as possible, judges that as many operators with safety helmets appear in the picture as possible according to the judgment, and controls the camera to carry out panoramic snapshot to leave a bottom when the system is adjusted to a proper position.
As a preferred embodiment of the present invention, the positioning of the person and the climbing carrier target in S2 specifically includes:
after a snapshot panoramic picture is collected through the S1 camera, the approximate directions of the positioning personnel and the climbing object are identified through the YOLO target detection algorithm model, in order to adapt to the computing power of the edge computing platform, model compression and acceleration are carried out on the YOLO target detection algorithm model, inference acceleration of the model is carried out through a specific neural network acceleration framework, and the real-time performance of the detection algorithm is guaranteed.
As a preferable scheme of the present invention, the determining of the position relationship between the vector and the person in S3 specifically includes:
after the positions of the climbing objects and the personnel in the panoramic picture are calculated through the personnel detection algorithm of S2, the algorithm model judges the position relation between the climbing objects and the characters in the target picture and judges whether the picture has climbing behaviors or not.
As a preferable scheme of the present invention, the step of controlling the placement control ball to obtain the local map of the climber in S4 specifically includes:
after the position of the climbing person in the panoramic picture is analyzed and identified through the S3 algorithm, the position relation between the coordinate frame of the climbing object, the coordinate frame of the climbing person and a target area frame preset by the camera is controlled through an ONVIF camera control protocol to move, zoom and zoom, and the camera is enlarged to a local area near the climbing person to take a snapshot.
As a preferred embodiment of the present invention, the segmentation of the person instance in S5 specifically includes:
the local pictures of the ascending personnel acquired through S4 may contain complex backgrounds, in order to reduce the influence of background factors on the safety belt identification performance and simultaneously perform subsequent image enhancement on the safety belt images on the human body, the personnel instance segmentation algorithm can realize pixel-level segmentation on the operators, eliminate the influence of the picture background, provide accurate positioning for the areas processed by the subsequent safety belt image enhancement algorithm, and reduce the calculation loss brought by the enhanced picture full-image.
As a preferable aspect of the present invention, the enhancing of the seat belt image in S6 specifically includes:
after the step of dividing the person example at S5, the possible existence region of the seat belt in the picture is determined. The shape and the color of the safety belt are possibly similar to the colors of outer clothing and a backpack of a person, so that the situation that the overall color of the whole person picture is consistent is caused.
As a preferable aspect of the present invention, the image quality evaluation in S7 specifically includes:
the image quality evaluation algorithm can filter the low-resolution and seriously-shielded video pictures captured by the cloth control ball, retain the high-quality safety belt pictures with clear pictures and no shielding, and provide the pictures with higher quality for the subsequent safety belt recognition algorithm.
As a preferable scheme of the present invention, the seat belt wearing identification in S8 specifically includes:
after a series of image processing of S5, S6 and S7, the seat belt detection is performed on the aerial worker by using a YOLO-based seat belt detection algorithm, in this step, the seat belt detection frame and the ascending person detection frame need to be matched, the matched seat belt can correspond to a person who wears the seat belt, and if the corresponding seat belt is not matched in the detection frame of a person, the person determines that dangerous behaviors of ascending the unbelted seat belt exist.
Compared with the prior art, the invention has the beneficial effects that:
1. according to the system, the intelligent snapshot of the actual operation scene is realized by controlling the movement and the zooming control of the camera, the recognition effect of the ascending behavior of the operator and the wearing of the safety belt can be improved, and the problem of low accuracy of the target detection algorithm under a long distance is solved.
2. In the invention, the image quality filtering module is added in front of the safety belt detection algorithm, so that a low-quality ball distribution and control video picture is filtered, and the detection effect of the safety belt detection algorithm is improved.
3. In the invention, the safety belt image enhancement module is added in front of the safety belt detection algorithm, and the shape and the color of the safety belt are highlighted by performing quality compensation and contrast correction on the image, so that the detection effect of the safety belt detection algorithm is improved.
4. In the invention, the system uses a personnel instance segmentation algorithm to separate pedestrians from the background and other objects in the picture, and only carries out image enhancement in the personnel area, thereby reducing the time required by the image enhancement algorithm.
Drawings
FIG. 1 is a schematic diagram of a system route frame structure according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, rather than all embodiments, and all other embodiments obtained by a person of ordinary skill in the art without any creative work based on the embodiments of the present invention belong to the protection scope of the present invention.
While several embodiments of the present invention will be described below in order to facilitate an understanding of the invention, with reference to the related description, the invention may be embodied in many different forms and is not limited to the embodiments described herein, but rather, these embodiments are provided so that this disclosure will be thorough and complete.
It will be understood that when an element is referred to as being "secured to" another element, it can be directly on the other element or intervening elements may also be present, that when an element is referred to as being "connected" to another element, it can be directly connected to the other element or intervening elements may also be present, and that the terms "vertical", "horizontal", "left", "right" and the like are used herein for descriptive purposes only.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs, and the terms used herein in the specification of the present invention are for the purpose of describing particular embodiments only and are not intended to limit the present invention, and the term "and/or" as used herein includes any and all combinations of one or more of the associated listed items.
Referring to fig. 1, the present invention provides a technical solution:
a portable cloth control ball-based identification system for a safety belt not tied during ascending is disclosed, which comprises the following specific operation procedures:
s1, carrying out panoramic snapshot;
s2, positioning the personnel and the climbing carrier target;
s3, judging the position relation between the carrier and the personnel;
s4, controlling the cloth control ball to obtain a local map of the person ascending a height;
s5, segmenting the personnel instance;
s6, enhancing the safety belt image;
s7, evaluating the image quality;
and S8, identifying the wearing of the safety belt.
Further, the panoramic capturing in S1 specifically includes:
the system can control the shooting angle and the focal length of the camera through an ONVIF protocol, ensures that an operation site is completely presented in a video picture as far as possible, judges that as many operators with safety helmets appear in the picture as possible according to the judgment, and controls the camera to carry out panoramic snapshot to leave a bottom when the system is adjusted to a proper position.
Further, the positioning of the personnel and the climbing carrier target in S2 specifically includes:
after a snapshot panoramic picture is collected through the S1 camera, the approximate directions of the positioning personnel and the climbing object are identified through the YOLO target detection algorithm model, in order to adapt to the computing power of the edge computing platform, model compression and acceleration are carried out on the YOLO target detection algorithm model, inference acceleration of the model is carried out through a specific neural network acceleration framework, and the real-time performance of the detection algorithm is guaranteed.
Further, the determining of the position relationship between the carrier and the person in S3 specifically includes:
after the positions of the climbing objects and the personnel in the panoramic picture are calculated through the personnel detection algorithm of S2, the algorithm model judges the position relation between the climbing objects and the characters in the target picture and judges whether the picture has climbing behaviors or not.
Further, the step of controlling the placement control ball to obtain the local map of the climber in step S4 specifically includes:
after the position of the climbing person in the panoramic picture is analyzed and identified through the S3 algorithm, the position relation between the coordinate frame of the climbing object, the coordinate frame of the climbing person and a target area frame preset by the camera is controlled through an ONVIF camera control protocol to move, zoom and zoom, and the camera is enlarged to a local area near the climbing person to take a snapshot.
Further, the segmentation of the person instance in S5 specifically includes:
the local pictures of the ascending personnel acquired through S4 may contain complex backgrounds, in order to reduce the influence of background factors on the safety belt identification performance and simultaneously perform subsequent image enhancement on the safety belt images on the human body, the personnel instance segmentation algorithm can realize pixel-level segmentation on the operators, eliminate the influence of the picture background, provide accurate positioning for the areas processed by the subsequent safety belt image enhancement algorithm, and reduce the calculation loss brought by the enhanced picture full-image.
Further, the enhancing of the seat belt image in S6 specifically includes:
after the step of dividing the person example at S5, the possible existence region of the seat belt in the picture is determined. The shape and the color of the safety belt are possibly similar to the colors of outer clothing and a backpack of a person, so that the situation that the overall color of the whole person picture is consistent is caused.
Further, the image quality evaluation in S7 specifically includes:
the image quality evaluation algorithm can filter the low-resolution and seriously-shielded video pictures captured by the cloth control ball, retain the high-quality safety belt pictures with clear pictures and no shielding, and provide the pictures with higher quality for the subsequent safety belt recognition algorithm.
Further, the seat belt wearing identification in S8 specifically includes:
after a series of image processing of S5, S6 and S7, the seat belt detection is performed on the aerial worker by using a YOLO-based seat belt detection algorithm, in this step, the seat belt detection frame and the ascending person detection frame need to be matched, the matched seat belt can correspond to a person who wears the seat belt, and if the corresponding seat belt is not matched in the detection frame of a person, the person determines that dangerous behaviors of ascending the unbelted seat belt exist.
The specific implementation case is as follows:
please refer to fig. 1:
the method comprises the following steps: carrying out panoramic snapshot;
the system can control the shooting angle and the focal length of the camera through an ONVIF protocol, ensures that an operation site is completely presented in a video picture as much as possible, judges that as many operators with safety helmets appear in the picture as possible, and controls the camera to carry out panoramic snapshot to leave a bottom when the system is adjusted to a proper position;
step two: positioning the personnel and the climbing carrier target;
after a snapshot panoramic picture is acquired by a camera in the first step, the approximate directions of positioning personnel and a climbing object are identified through a YOLO target detection algorithm model, in order to adapt to the computing power of an edge computing platform, model compression and acceleration are carried out on the YOLO target detection algorithm model, inference acceleration of the model is carried out through a specific neural network acceleration frame, and the real-time performance of the detection algorithm is guaranteed;
step three: judging the position relation between the carrier and the personnel;
after the positions of the climbing objects and the personnel in the panoramic picture are calculated through the personnel detection algorithm in the second step, the algorithm model judges the position relation of the climbing objects and the characters in the target picture and judges whether the picture has climbing behaviors or not;
step four: controlling a distribution control ball to obtain a local map of the person ascending a height;
after the position of a climbing person in the panoramic picture is analyzed and identified through the third algorithm, the position relation between a coordinate frame of a climbing object, the coordinate frame of the climbing person and a target area frame preset by the camera is controlled through an ONVIF camera control protocol to move, zoom and zoom, and the camera is enlarged to a local area near the climbing person to take a snapshot;
step five: dividing a person example;
the local pictures of the ascending personnel obtained in the fourth step may contain complex backgrounds, in order to reduce the influence of background factors on the safety belt identification performance, and meanwhile, the subsequent image enhancement is carried out on the safety belt images on the human body, the personnel instance segmentation algorithm can realize pixel-level segmentation on the operators, the influence of the picture background is eliminated, accurate positioning is provided for the areas processed by the subsequent safety belt image enhancement algorithm, and the calculation loss caused by the enhancement of the whole picture is reduced;
step six: enhancing the safety belt image;
after the step of five person example segmentation steps, the possible existing area of the safety belt in the picture is determined. The shape and the color of the safety belt are probably similar to the colors of outer clothing and a backpack of a person, so that the situation that the overall color of the overall person picture is consistent is caused, the image contrast of an operator area is increased through a safety belt image enhancement algorithm in the step, the appearance and the color characteristics of the safety belt are highlighted, and the identification performance of a subsequent identification algorithm on the safety belt is improved;
step seven: evaluating the image quality;
the image quality evaluation algorithm can filter the low-resolution and seriously-shielded video pictures captured by the cloth control ball, retain the high-quality safety belt pictures with clear pictures and no shielding, and provide pictures with higher quality for the subsequent safety belt recognition algorithm;
step eight: identifying the wearing of the safety belt;
after a series of image processing of the fifth step, the sixth step and the seventh step, safety belt detection is carried out on the high-altitude operation personnel by using a safety belt detection algorithm based on YOLO, in the step, the safety belt detection frame and the ascending personnel detection frame are matched, the matched safety belt can correspond to a person who wears the safety belt, and if the corresponding safety belt is not matched in the detection frame of a certain person, the person judges that dangerous behaviors of not fastening the safety belt at the ascending height exist.
The key technical points of the invention are as follows:
1. intelligent snapshot technology for operation scene based on zoom switching
In order to solve the problem of low accuracy of target detection and behavior recognition algorithm under a long distance, the system realizes intelligent snapshot of an actual operation scene by controlling the movement and zooming control of the camera, and can improve the recognition effect of the ascending behavior and the escalator behavior of an operator.
Detailed technical description: according to the actual situation of an operation field, the focal length of the camera can be adjusted through an ONVIF camera control protocol, different visual angles (a macroscopic visual angle and a microscopic visual angle) are switched to identify ascending operation behaviors and escalator behaviors, under the macroscopic visual angle, whether personnel are conducting ascending operation or not is identified through judging the space position between the personnel and the escalator, under the microscopic visual angle, whether the personnel are conducting ascending operation or not is identified through analyzing the posture of the personnel, and abnormal behaviors of the personnel in a picture are captured.
2. Intelligent personnel instance segmentation technology in operation scene based on deep learning
The pedestrian instance segmentation algorithm based on deep learning can realize the segmentation of the pixel level of the human. The system separates people from the background through a pedestrian instance segmentation algorithm, reduces interference and influence of sky, electric wires, equipment and the like on identification, can improve the performance of a pedestrian detection algorithm under different visual angles (a macroscopic visual angle and a microscopic visual angle), and can serve as one of filtering conditions of non-human interference objects.
3. Safety belt image enhancement technology for climbing operator based on compensation correction
For most of images captured by distributed ball control, the whole image is darker or brighter, because the range of the gray value of the image is smaller, that is, the contrast is low, for the image with lower contrast, the image can be preprocessed (image enhanced) to a certain degree before the target detection algorithm and the example segmentation algorithm are carried out to increase the contrast of the image, and the identification performance is improved.
The safety belt image enhancement technology used by the system is completed based on the image enhancement technology of gain compensation and gamma conversion, and can perform definition enhancement and sharpening on the climbing operation personnel and the safety belt images thereof, so that the detailed effect of the safety belt worn on the body of the personnel is highlighted.
4. Image filtering technology of climbing operator based on brightness, definition and shielding degree
The filtering of the image to be recognized is an important link for ensuring that the system does not report by mistake or report by mistake, the climbing operator recognizes that the brightness of the image is abnormal due to environmental factors (sunlight irradiation, rainy weather and the like), the definition of the image captured in the zooming process of the camera or under the condition of incorrect zooming is reduced, and the recognition precision is reduced due to the shielding of objects such as electric wires and the like in an operation scene.
The system designs an image quality filtering algorithm module to judge the brightness, the definition and the sheltered degree of personnel, filters low-quality images, avoids false reports and false reports, and is an important link of the whole system.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (9)
1. A portable cloth ball control-based identification system for a climbing unbelted safety belt is characterized by comprising the following specific operation procedures:
s1, carrying out panoramic snapshot;
s2, positioning the personnel and the climbing carrier target;
s3, judging the position relation between the carrier and the personnel;
s4, controlling the cloth control ball to obtain a local map of the person ascending a height;
s5, segmenting the personnel instance;
s6, enhancing the safety belt image;
s7, evaluating the image quality;
and S8, identifying the wearing of the safety belt.
2. The portable ball-on-cloth based hiking unbelted belt identification system of claim 1, wherein: the panoramic snapshot in S1 specifically includes:
the system can control the shooting angle and the focal length of the camera through an ONVIF protocol, ensures that an operation site is completely presented in a video picture as far as possible, judges that as many operators with safety helmets appear in the picture as possible according to the judgment, and controls the camera to carry out panoramic snapshot to leave a bottom when the system is adjusted to a proper position.
3. The portable ball-on-cloth based hiking unbelted belt identification system of claim 1, wherein: the positioning of the personnel and the climbing carrier target in the S2 specifically comprises the following steps:
after a snapshot panoramic picture is collected through the S1 camera, the approximate directions of the positioning personnel and the climbing object are identified through the YOLO target detection algorithm model, in order to adapt to the computing power of the edge computing platform, model compression and acceleration are carried out on the YOLO target detection algorithm model, inference acceleration of the model is carried out through a specific neural network acceleration framework, and the real-time performance of the detection algorithm is guaranteed.
4. The portable ball-on-cloth based hiking unbelted belt identification system of claim 1, wherein: the judging of the position relationship between the carrier and the personnel in the S3 specifically comprises the following steps:
after the positions of the climbing objects and the personnel in the panoramic picture are calculated through the personnel detection algorithm of S2, the algorithm model judges the position relation between the climbing objects and the characters in the target picture and judges whether the picture has climbing behaviors or not.
5. The portable ball-on-cloth based hiking unbelted belt identification system of claim 1, wherein: the step S4 of controlling the placement control ball to obtain the local map of the climber specifically includes:
after the position of the climbing person in the panoramic picture is analyzed and identified through the S3 algorithm, the position relation between the coordinate frame of the climbing object, the coordinate frame of the climbing person and a target area frame preset by the camera is controlled through an ONVIF camera control protocol to move, zoom and zoom, and the camera is enlarged to a local area near the climbing person to take a snapshot.
6. The portable ball-on-cloth based hiking unbelted belt identification system of claim 1, wherein: the segmentation of the person instance in S5 specifically includes:
the local pictures of the ascending personnel acquired through S4 may contain complex backgrounds, in order to reduce the influence of background factors on the safety belt identification performance and simultaneously perform subsequent image enhancement on the safety belt images on the human body, the personnel instance segmentation algorithm can realize pixel-level segmentation on the operators, eliminate the influence of the picture background, provide accurate positioning for the areas processed by the subsequent safety belt image enhancement algorithm, and reduce the calculation loss brought by the enhanced picture full-image.
7. The portable ball-on-cloth based hiking unbelted belt identification system of claim 1, wherein: the enhancing of the seat belt image in S6 specifically includes:
after the step of dividing the person example at S5, the possible existence region of the seat belt in the picture is determined. The shape and the color of the safety belt are possibly similar to the colors of outer clothing and a backpack of a person, so that the situation that the overall color of the whole person picture is consistent is caused.
8. The portable ball-on-cloth based hiking unbelted belt identification system of claim 1, wherein: the image quality evaluation in S7 specifically includes:
the image quality evaluation algorithm can filter the low-resolution and seriously-shielded video pictures captured by the cloth control ball, retain the high-quality safety belt pictures with clear pictures and no shielding, and provide the pictures with higher quality for the subsequent safety belt recognition algorithm.
9. The portable ball-on-cloth based hiking unbelted belt identification system of claim 1, wherein: the step S8 of recognizing the wearing of the seat belt specifically includes:
after a series of image processing of S5, S6 and S7, the seat belt detection is performed on the aerial worker by using a YOLO-based seat belt detection algorithm, in this step, the seat belt detection frame and the ascending person detection frame need to be matched, the matched seat belt can correspond to a person who wears the seat belt, and if the corresponding seat belt is not matched in the detection frame of a person, the person determines that dangerous behaviors of ascending the unbelted seat belt exist.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111615156.7A CN114359761A (en) | 2021-12-27 | 2021-12-27 | Climbing unbuckled safety belt recognition system based on portable cloth ball control |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111615156.7A CN114359761A (en) | 2021-12-27 | 2021-12-27 | Climbing unbuckled safety belt recognition system based on portable cloth ball control |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114359761A true CN114359761A (en) | 2022-04-15 |
Family
ID=81103918
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111615156.7A Pending CN114359761A (en) | 2021-12-27 | 2021-12-27 | Climbing unbuckled safety belt recognition system based on portable cloth ball control |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114359761A (en) |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107862282A (en) * | 2017-11-07 | 2018-03-30 | 深圳市金城保密技术有限公司 | A kind of finger vena identification and safety certifying method and its terminal and system |
CN109218695A (en) * | 2017-06-30 | 2019-01-15 | 中国电信股份有限公司 | Video image enhancing method, device, analysis system and storage medium |
CN110852183A (en) * | 2019-10-21 | 2020-02-28 | 广州大学 | Method, system, device and storage medium for identifying person without wearing safety helmet |
WO2020056677A1 (en) * | 2018-09-20 | 2020-03-26 | 中建科技有限公司深圳分公司 | Violation detection method, system, and device for building construction site |
AU2020100705A4 (en) * | 2020-05-05 | 2020-06-18 | Chang, Jiaying Miss | A helmet detection method with lightweight backbone based on yolov3 network |
CN111898514A (en) * | 2020-07-24 | 2020-11-06 | 燕山大学 | Multi-target visual supervision method based on target detection and action recognition |
CN113076825A (en) * | 2021-03-19 | 2021-07-06 | 云南电网有限责任公司西双版纳供电局 | Transformer substation worker climbing safety monitoring method |
CN113450318A (en) * | 2021-06-11 | 2021-09-28 | 东华大学 | Primary power distribution system porcelain insulator detection method based on unmanned aerial vehicle autonomous vision |
-
2021
- 2021-12-27 CN CN202111615156.7A patent/CN114359761A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109218695A (en) * | 2017-06-30 | 2019-01-15 | 中国电信股份有限公司 | Video image enhancing method, device, analysis system and storage medium |
CN107862282A (en) * | 2017-11-07 | 2018-03-30 | 深圳市金城保密技术有限公司 | A kind of finger vena identification and safety certifying method and its terminal and system |
WO2020056677A1 (en) * | 2018-09-20 | 2020-03-26 | 中建科技有限公司深圳分公司 | Violation detection method, system, and device for building construction site |
CN110852183A (en) * | 2019-10-21 | 2020-02-28 | 广州大学 | Method, system, device and storage medium for identifying person without wearing safety helmet |
AU2020100705A4 (en) * | 2020-05-05 | 2020-06-18 | Chang, Jiaying Miss | A helmet detection method with lightweight backbone based on yolov3 network |
CN111898514A (en) * | 2020-07-24 | 2020-11-06 | 燕山大学 | Multi-target visual supervision method based on target detection and action recognition |
CN113076825A (en) * | 2021-03-19 | 2021-07-06 | 云南电网有限责任公司西双版纳供电局 | Transformer substation worker climbing safety monitoring method |
CN113450318A (en) * | 2021-06-11 | 2021-09-28 | 东华大学 | Primary power distribution system porcelain insulator detection method based on unmanned aerial vehicle autonomous vision |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110070033B (en) | Method for detecting wearing state of safety helmet in dangerous working area in power field | |
CN111126325B (en) | Intelligent personnel security identification statistical method based on video | |
CN108062542B (en) | Method for detecting shielded human face | |
WO2017028587A1 (en) | Vehicle monitoring method and apparatus, processor, and image acquisition device | |
CN103942539B (en) | A kind of oval accurate high efficiency extraction of head part and masking method for detecting human face | |
CN105160297B (en) | Masked man's event automatic detection method based on features of skin colors | |
CN104166861A (en) | Pedestrian detection method | |
CN108205649B (en) | Method and device for recognizing state of driver for calling and answering | |
CN109672863A (en) | A kind of construction personnel's safety equipment intelligent monitoring method based on image recognition | |
CN101930535A (en) | Human face detection and tracking device | |
CN106339657B (en) | Crop straw burning monitoring method based on monitor video, device | |
CN112396658A (en) | Indoor personnel positioning method and positioning system based on video | |
CN102542246A (en) | Abnormal face detection method for ATM (Automatic Teller Machine) | |
CN111368727B (en) | Dressing detection method, storage medium, system and device for inspection personnel in power distribution room | |
CN113903081A (en) | Visual identification artificial intelligence alarm method and device for images of hydraulic power plant | |
CN112183472A (en) | Method for detecting whether test field personnel wear work clothes or not based on improved RetinaNet | |
CN107368786B (en) | Machine vision-based passenger passing handrail detection method | |
CN112487891B (en) | Visual intelligent dynamic identification model construction method applied to electric power operation site | |
CN112101260B (en) | Method, device, equipment and storage medium for identifying safety belt of operator | |
CN105069816B (en) | A kind of method and system of inlet and outlet people flow rate statistical | |
CN108182381A (en) | Escalator occupant detection algorithm based on quick Adaboost training algorithms | |
CN115035088A (en) | Helmet wearing detection method based on yolov5 and posture estimation | |
CN114187543A (en) | Safety belt detection method and system in high-altitude power operation scene | |
CN103578121B (en) | Method for testing motion based on shared Gauss model under disturbed motion environment | |
KR102171384B1 (en) | Object recognition system and method using image correction filter |
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 |