Method for recognizing unsafe behaviors of coal mine workers based on human body postures

Abstract

The invention discloses a method for recognizing unsafe behaviors of coal mine workers based on human body postures. The method comprises the following steps: firstly, presetting a plurality of unsafebehaviors of coal mine workers, acquiring video information of the unsafe behaviors, determining positions and categories of human body joints in the video information through a human body detector,and calculating affinity vector fields of each group of adjacent human body joints in the unsafe behaviors to serve as threshold values; secondly, acquiring a working image of a coal mine worker through a camera, and predicting the position and category of a human body joint in the working image by using a regression method; and finally, calculating an affinity vector field of adjacent human bodyjoints in the working image, comparing the affinity vector field with the threshold value, judging that the worker behavior violates rules if the affinity vector field and the threshold value are within a certain difference rang, and giving an alarm; and if the difference is not within the range, judging the behavior to be not illegal. An alarm can be given on the three-violation behavior of coalmine workers, and the occurrence of the three-violation behavior can be reduced.

Description

technical field [0001] The invention relates to the technical field of safety monitoring, in particular to a method for identifying unsafe behaviors of coal mine workers based on human body gestures. Background technique [0002] In coal mine production, the accidents caused by irregular operations, illegal command and violation of labor discipline reflect the inevitable result of the regular behavior of personnel. The main body of the "three violations" is people, and the phenomenon of "three violations" by on-site management personnel and operating workers occurs from time to time, and repeated prohibitions continue, which seriously threaten the safe production of mines and the lives of employees. [0003] At present, the supervision of "three violations" in the coal mining industry is mainly through video surveillance and on-site manual supervision. There are also a few mining areas that have introduced automatic monitoring systems and established artificial intelligence ...

AI Technical Summary

Problems solved by technology

[0004] 1. The efficiency of ordinary target recognition algorithms needs to be improved. The underground light is dark and the situation is complicated. In order to meet the needs of computer vision algorithms, more expensive high-definition cameras must be reinstalled

[0005] 2. In the darker light or when the helmet is not a conventional type, the helmet cannot be recognized

[0006] 3. It is difficult to recognize long-distance fuzzy faces, and it is difficult to confirm the identity of violators

[0007] 4. The process of analysis and identification of posture structured data in coal miner posture analysis results is easily disturbed by underground environment, light, occlusion and other factors, and there are false positives in the "three violations" behavior

[0008] 5. In order to reduce overfitting, ordinary deep learning models require a large amount of data for model training. It is difficult to provide enough data sets in the conventional coal mine production process. High rate of "three violations" behavior artificial intelligence model

Images