Target crowd tracking and monitoring method based on limited range scene

Abstract

The invention discloses a target crowd tracking and monitoring method based on a limited range scene, and the method comprises the steps: S1, designing and achieving a whole set of camera placement and collection device based on a limited visual field range, so as to meet the demands of a target crowd activity scene and the full coverage of a visual field; S2, modifying the size and proportion ofan anchor frame of a detection network model according to the body type and face type proportion of the target crowd, and modifying a loss function of the detection network to improve the detection accuracy; S3, recognizing the human face by adopting a specific recognition network while detecting the human face, so as to improve the detection and recognition precision while not reducing the speedunder the condition that the detection network and the human face recognition network are used at the same time; and S4, improving a multi-target tracking monitoring model, and improving the model adaptability of target crowd tracking monitoring through extraction of specific features and pruning of a specific scene. According to the design scheme, in a visual scheme of nursing the target crowd and preventing accidents, the monitoring accuracy and the positioning speed are improved while no leakage and real-time performance are realized.

Description

technical field [0001] The invention relates to a method for tracking and monitoring a target group based on a scene within a limited range, and belongs to the field of target identification, tracking and monitoring. Background technique [0002] The breakthroughs made by artificial intelligence technology in the field of computer vision have ushered in a round of explosions in artificial intelligence, and deep learning is the key technology to achieve these breakthroughs. Among them, the image classification and recognition technology based on deep convolutional network has exceeded the accuracy of human eyes, and the image recognition technology based on deep neural network has reached an accuracy rate of 95%. In some limited scenarios such as nurseries, nursing homes, hospitals, and nursing homes, the target population needs real-time monitoring and assistance. However, under the existing technical conditions, there are defects such as untimely assistance and difficulty i...

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to generate an algorithm specially aimed at the combination of face detection and face recognition and tracking monitoring of the target group through the improvement of the existing target detection and tracking monitoring algorithm, which can effectively avoid occlusion and ID instability Changes and other issues

Images