Elbow bending behavior detection method based on human body key points

Abstract

The invention discloses an elbow bending behavior detection method based on human body key points, and the method comprises the steps: inputting a to-be-detected video or picture, and carrying out the positioning of a person target; obtaining a high-quality person target candidate frame by using a regional multi-person pose estimation frame (RMPE), and obtaining coordinate information and confidence scores of 17 human body key points; threshold processing is carried out on the confidence score, and key point coordinates with low detection precision are removed; in combination with human body key point coordinate information, a three-point positioning method is provided, an elbow bending behavior probability value is obtained according to an angle range and a distance ratio value between key points, and a small target detection area is positioned; detecting a small target object by using transfer learning and a target detection algorithm to obtain a confidence score of a target category; and combining the elbow flexion behavior probability value and the target category confidence score to realize detection of the elbow flexion behavior of the human body. According to the method, human elbow bending behavior detection in a complex scene can be realized, and potential safety hazard behaviors can be detected in time and early warning is performed.

Description

technical field [0001] The invention belongs to the field of computer graphics and image processing, and relates to an elbow-bending behavior detection method based on key points of a human body. Background technique [0002] With the continuous advancement of artificial intelligence technology, the application of computer vision in various aspects is becoming more and more extensive, especially in the aspect of intelligent identification and analysis of video surveillance. Traditional manual monitoring camera images consume a lot of manpower and material resources, are inefficient and prone to misjudgments and omissions, and the emergence of intelligent video monitoring allows computers to automatically monitor violations and alarms, which is of great significance. At present, some progress has been made in human body key point detection and behavior analysis. These methods have a good recognition effect for a single scene or specific behavior. Two-dimensional human key po...

AI Technical Summary

Problems solved by technology

For other complex places such as offshore oil operation platforms, the detection background is complex, there are many interference objects, and the video picture is blurred. It is difficult to obtain a high recognition accuracy rate, which increases the difficulty of target detection and video analysis, and detects small target objects in the picture become more difficult

Images