CN113392708A - Safety belt detection method - Google Patents
Safety belt detection method Download PDFInfo
- Publication number
- CN113392708A CN113392708A CN202110525232.9A CN202110525232A CN113392708A CN 113392708 A CN113392708 A CN 113392708A CN 202110525232 A CN202110525232 A CN 202110525232A CN 113392708 A CN113392708 A CN 113392708A
- Authority
- CN
- China
- Prior art keywords
- safety belt
- safety
- detection method
- video frame
- seat 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
- 238000001514 detection method Methods 0.000 title claims abstract description 32
- 230000011218 segmentation Effects 0.000 claims description 9
- 238000000034 method Methods 0.000 claims description 7
- 238000001914 filtration Methods 0.000 abstract description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
Abstract
The invention provides a safety belt detection method, which comprises the steps of firstly obtaining pictures through video streaming, then segmenting a target object through a detection network, filtering unclear or shielded situations and the like through a classifier, then segmenting the network, carrying out fine-grained classification, and further determining whether a safety belt is worn correctly through reasonable logic judgment (whether the safety belt belongs to high hanging and low hanging). The invention can provide a reliable and efficient safety belt detection method, can solve the objective problems of shielding, unclear property and the like, and reduces the false alarm rate; meanwhile, due to the fact that occlusion is filtered out, the picture is not clear, time is not spent on the picture which cannot be judged by human eyes, and project efficiency is further improved.
Description
Technical Field
The invention relates to the field of visual algorithms, in particular to a safety belt detection method.
Background
In high-altitude operation, due to the complex scene and the large number of people, the situation of false alarm of safety belt detection occurs. In order to remove the unclear picture, image quality classification is performed, and in order to remove the stoop and the side, human posture estimation is applied. At present, yolov5s, which has advanced performance and speed, is used more frequently in the detection. In the segmentation algorithm, aiming at a small target and considering the speed, the structure of using the FPNNetFPN is simple, and can be summarized as follows: feature extraction, up-sampling, feature fusion and multi-scale feature output. The FPN inputs pictures with any size and outputs feature maps with various scales. The overall network structure of the FPN is divided into two parts, bottom-up and top-down. The bottom-up is a feature extraction process, namely an Encoder part, the features of the deepest layer are sampled from the top down to the resolution corresponding to the bottom-up output through layer-by-layer upsampling, and the resolution is fused with the sampled resolution to output a feature map. The prior art has the defects of misinformation and low efficiency.
Therefore, a reliable and efficient safety belt detection method is needed to solve the safety belt detection problem under the conditions that the ambiguity is unclear and people block the side.
Disclosure of Invention
The invention aims to solve the problem of false alarm of safety belt detection caused by complicated scene and more people or the conditions of fuzzy and unclear people, side shielding and the like in high-altitude operation.
In order to achieve the object, the present invention provides a seat belt detection method for performing robust seat belt detection in a case where a person is unclear and blocks a side body. The method comprises the steps of obtaining pictures through video streaming, segmenting a target object through a detection network, filtering unclear or sheltered objects through a classifier, classifying fine grit through a segmentation network, and determining whether a safety belt is worn correctly through reasonable logic judgment (whether high hanging and low hanging are needed or not). The safety belt is hung at a position higher than a person to stand, and the working position of the person is lower than the position of the safety belt hanger. In this manner, the combined forces of the safety belt, safety line and metal fitting can hold the person in case of a fall, reducing the actual impact distance or preventing it from falling.
Firstly, acquiring a video frame, then carrying out target detection (human), then entering a classifier, and entering a definition step after the classifier step; clearly defined positive and negative examples; the positive sample comprises one or more of the conditions that the upper half of the body is not shielded, the clear front and back sides of the person are shielded, the safety belt is totally clear, and the front and back sides of the person are shielded; the negative examples include: leaning on the body; stoop and squat; the upper half body completely covers or completely shields the unclear front and back sides of the human body; most of the safety belts are shielded or most of the unclear front and back sides of the people; the upper half of the body is shielded, and the safety belt is partially shielded or partially obscured in one or more of the front and back of a person.
If the definition is not clear, returning to the first step to obtain the video frame again; if the definition is clear, the sample is segmented into adults, safety belts or safety ropes.
After the network splitting step, it is determined whether a seat belt or a safety line is detected. If the safety belt or the safety rope is detected, whether the vehicle is used for hanging high or hanging low is further judged. If the high hanging low use is available, returning to the first step to reacquire the video frame, and if the high hanging low use is not available, entering the step of the connected domain near the upper half body safety belt.
And after entering the step of the connected domain near the upper half body safety belt, judging whether the connected domains are all the pixel points of people, if not, returning to the first step to acquire the video frame again, if so, giving an alarm to the safety belt, and then returning to the first step to acquire the video frame again.
Furthermore, the safety belt detection method can also directly enter a segmentation network after the target detection step, and segment the sample into adults, safety belts or safety ropes.
After the network segmentation step, judging the detected safety belt or safety rope, if so, performing a step of judging whether the safety belt or safety rope is used for high hanging or low hanging, and if not, entering a classifier; if the video frame is used in a high hanging low mode, returning to the first step, and obtaining the video frame again, and if the video frame is not used in a high hanging low mode, entering a classifier.
After the classifier step, judging whether the upper half body is a work clothes, if not, switching to the first step, and acquiring the video frame again; if so, carrying out safety belt alarm, and then turning to the first step to acquire the video frame again.
The method has the advantages that objective problems of shielding, unclear effect and the like are solved, and the false alarm rate is reduced; meanwhile, due to the fact that occlusion is filtered out, the picture is not clear, time is not spent on the picture which cannot be judged by human eyes, and project efficiency is further improved.
In order to make the aforementioned and other objects, features and advantages of the present invention comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
Fig. 1 is a flowchart of a seat belt detection method of the present invention.
Fig. 2 is a flow chart of another seat belt detection method of the present invention.
Detailed Description
The technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention. It is to be understood that the described embodiments are merely exemplary of the invention, and not restrictive of the full scope of the invention. The components of embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present invention without making any creative effort, shall fall within the protection scope of the present invention.
The present invention is described in further detail below to enable those skilled in the art to practice the invention with reference to the description.
Referring to fig. 1, fig. 1 is a flowchart of a seat belt detection method according to the present invention. In fig. 1, a video frame is first acquired, and then target detection is performed to detect whether a target is a human or not. And then the program enters a classifier to perform definition clear judgment. If the definition is clear, the program enters a segmentation step to segment the target into a person, a safety belt and a safety rope; if the definition is not clear, the procedure is returned to the beginning, and the video frame is acquired again. After the target is divided, judging whether the detected target is a safety belt or a safety rope, and if the detected target is the safety belt or the safety rope, further judging whether the target is used for hanging high or hanging low; if not, the upper body harness is entered into the connected area near the upper body harness. If the high-hang low-use is adopted, the return program starts to acquire the video frame again, and if the high-hang low-use is not adopted, the communication domain near the upper half body safety belt is entered. After entering a connected domain near the upper half body safety belt, starting to judge whether the connected domains are all human pixel points, if not, returning to a program to start to acquire video frames again; if yes, a safety belt alarm is carried out, and the program returns to start to acquire the video frames again.
The definition in fig. 1 clearly divides into positive and negative examples. The positive samples are divided into two cases: (1) the upper half body is not blocked, and the front and back sides of the human body are clear; (2) the upper half of the body is shielded, and the safety belt is completely clear and is positioned on the front side and the back side of the body. The negative examples are divided into five cases: (1) leaning on the body; (2) stooping and squatting; (3) the upper half body completely covers or completely shields the unclear front and back sides of the human body; (4) most of the safety belts are shielded or most of the unclear front and back sides of the people; (5) the upper half of the body is shielded, and the safety belt is partially shielded or partially obscured to the front and back of the person.
Referring to fig. 2, fig. 2 is a flowchart illustrating another seat belt detecting method according to the present invention. In fig. 2, unlike the embodiment of fig. 1, in the embodiment of fig. 2, the program starts to acquire a video frame, performs object detection on a human object, and directly performs segmentation after detecting the object person, so as to segment the object into a human, a seat belt and a safety rope. Then judging whether the detected safety belt and the detected safety rope are used, and if the detected safety belt and the detected safety rope are used, further judging whether the detected safety belt and the detected safety rope are used for hanging high or hanging low; if it is not a seat belt or safety line, the program enters the sorter. If the video frame is judged to be used in a high hang-up state and a low hang-up state, returning to the program to start to acquire the video frame again; and if the data is not used for high hanging and low hanging, entering a classifier. After entering the classifier, judging whether the upper half body is the worker clothes, if not, returning to the program to start to acquire the video stream again; if the upper half body is a work clothes, a safety belt alarm is sent out, and the program returns to start to acquire the video stream again.
While embodiments of the invention have been disclosed above, it is not intended to be limited to the uses set forth in the specification and examples. It can be applied to all kinds of fields suitable for the present invention. Additional modifications will readily occur to those skilled in the art. Therefore, the invention is not to be limited to the specific details and illustrations shown and described herein, without departing from the general concept defined by the claims and their equivalents.
Claims (8)
1. The safety belt detection method comprises the following steps:
firstly, pictures are obtained through video streaming, a target object is segmented through a detection network, then the unclear or shielding condition and the like are filtered through a classifier, fine-grained classification is carried out through the segmentation network, and then whether the safety belt is worn correctly is determined through judging whether the safety belt belongs to a high hanging position or a low hanging position.
2. The seat belt detection method according to claim 1, characterized by further comprising, after the classifier step, a definition step; clearly defined positive and negative examples; the positive sample comprises one or more of the conditions that the upper half of the body is not shielded, the clear front and back sides of the person are shielded, the safety belt is totally clear, and the front and back sides of the person are shielded; the negative examples include: leaning on the body; stoop and squat; the upper half body completely covers or completely shields the unclear front and back sides of the human body; most of the safety belts are shielded or most of the unclear front and back sides of the people; the upper half of the body is shielded, and the safety belt is partially shielded or partially obscured in one or more of the front and back of a person.
3. The seat belt detection method of claim 1, wherein the segmentation network segments the sample into an adult, a seat belt, and a seat belt tether, and further comprising determining whether a seat belt or a seat belt tether is detected after the step of segmenting the network.
4. The safety belt detection method according to claim 3, characterized in that after the safety belt and safety rope detection step, the method further comprises the steps of judging whether the safety belt and safety rope are used in a high hanging state or not, and if so, returning to the step of acquiring the video frame; if not, then enter the connected domain near the upper body harness.
5. The safety belt detection method according to claim 4, characterized in that after the step of the connected domain near the upper half body safety belt, the method further comprises the steps of judging whether the connected domains are all pixel points of people, if not, returning to the first step, and reacquiring the video frame; if yes, a safety belt alarm is carried out, and then the first step is returned to, and the video frame is obtained again.
6. The seat belt detection method of claim 1, wherein the target detection step is directly followed by entering a segmentation network to segment the sample into adults, seat belts, or seat belts.
7. The seat belt detecting method according to claim 6, wherein after the network splitting step, the detected seat belt or seat belt rope is judged, if yes, a step of judging high-hanging low is performed, and if not, the step enters a classifier;
if the video frame is used in a high hanging low mode, returning to the first step, and obtaining the video frame again, and if the video frame is not used in a high hanging low mode, entering a classifier.
8. The seat belt detection method according to claim 7, characterized in that after the step of classifying, it is determined whether the upper body is a work suit, and if not, the step proceeds to a first step to reacquire a video frame; if so, carrying out safety belt alarm, and then turning to the first step to acquire the video frame again.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110525232.9A CN113392708A (en) | 2021-05-13 | 2021-05-13 | Safety belt detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110525232.9A CN113392708A (en) | 2021-05-13 | 2021-05-13 | Safety belt detection method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113392708A true CN113392708A (en) | 2021-09-14 |
Family
ID=77617881
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110525232.9A Pending CN113392708A (en) | 2021-05-13 | 2021-05-13 | Safety belt detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113392708A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114155492A (en) * | 2021-12-09 | 2022-03-08 | 华电宁夏灵武发电有限公司 | High-altitude operation safety belt hanging rope high-hanging low-hanging use identification method and device and electronic equipment |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105488453A (en) * | 2015-11-30 | 2016-04-13 | 杭州全实鹰科技有限公司 | Detection identification method of no-seat-belt-fastening behavior of driver based on image processing |
| CN110084184A (en) * | 2019-04-25 | 2019-08-02 | 浙江吉利控股集团有限公司 | A kind of safety belt based on image processing techniques is not detection system and method |
| CN111414499A (en) * | 2020-05-08 | 2020-07-14 | 刘如意 | Operation personnel safety belt wearing detection system based on block chain and BIM |
| CN112215138A (en) * | 2020-10-12 | 2021-01-12 | 中国石油大学(华东) | Deep learning-based violation detection method for low hanging height of safety belt |
| CN112257580A (en) * | 2020-10-21 | 2021-01-22 | 中国石油大学(华东) | Human body key point positioning detection method based on deep learning |
-
2021
- 2021-05-13 CN CN202110525232.9A patent/CN113392708A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105488453A (en) * | 2015-11-30 | 2016-04-13 | 杭州全实鹰科技有限公司 | Detection identification method of no-seat-belt-fastening behavior of driver based on image processing |
| CN110084184A (en) * | 2019-04-25 | 2019-08-02 | 浙江吉利控股集团有限公司 | A kind of safety belt based on image processing techniques is not detection system and method |
| CN111414499A (en) * | 2020-05-08 | 2020-07-14 | 刘如意 | Operation personnel safety belt wearing detection system based on block chain and BIM |
| CN112215138A (en) * | 2020-10-12 | 2021-01-12 | 中国石油大学(华东) | Deep learning-based violation detection method for low hanging height of safety belt |
| CN112257580A (en) * | 2020-10-21 | 2021-01-22 | 中国石油大学(华东) | Human body key point positioning detection method based on deep learning |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114155492A (en) * | 2021-12-09 | 2022-03-08 | 华电宁夏灵武发电有限公司 | High-altitude operation safety belt hanging rope high-hanging low-hanging use identification method and device and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110532976B (en) | Fatigue driving detection method and system based on machine learning and multi-feature fusion | |
| CN107563347B (en) | Passenger flow counting method and device based on TOF camera | |
| CN101739550B (en) | Method and system for detecting moving objects | |
| CN104601965B (en) | Camera occlusion detection method | |
| CN106056079B (en) | A kind of occlusion detection method of image capture device and human face five-sense-organ | |
| TW201120814A (en) | Method for determining if an input image is a foggy image, method for determining a foggy level of an input image and cleaning method for foggy images | |
| CN105760847B (en) | A kind of visible detection method of pair of helmet of motorcycle driver wear condition | |
| CN107911663A (en) | A kind of elevator passenger hazardous act intelligent recognition early warning system based on Computer Vision Detection | |
| CN103839085A (en) | Train carriage abnormal crowd density detection method | |
| CN103400111A (en) | Method for detecting fire accident on expressway or in tunnel based on video detection technology | |
| CN111616718B (en) | Method and system for detecting fatigue state of driver based on attitude characteristics | |
| CN103729858A (en) | Method for detecting article left over in video monitoring system | |
| CN110009650B (en) | Escalator handrail boundary area border crossing detection method and system | |
| CN101916372A (en) | Video-based smoke detection device and method according to multi-feature fusion | |
| CN113392708A (en) | Safety belt detection method | |
| CN114596329A (en) | Gas image enhancement and gas leakage detection method and system | |
| CN103400113A (en) | Method for detecting pedestrian on expressway or in tunnel based on image processing | |
| CN110913209B (en) | Camera shielding detection method and device, electronic equipment and monitoring system | |
| CN107194946A (en) | A kind of infrared obvious object detection method based on FPGA | |
| CN103021179A (en) | Real-time monitoring video based safety belt detection method | |
| KR20090044957A (en) | Theft and left baggage survellance system and meothod thereof | |
| CN116703925B (en) | Bearing defect detection method and device, electronic equipment and storage medium | |
| JP2010003116A (en) | Object deciding device and program | |
| Grove et al. | Colour based object tracking | |
| CN110407052B (en) | Method and system for detecting violent movement behaviors in elevator |
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 |