Target detection method and system, computer equipment and machine readable medium

Abstract

The invention provides a target detection method and system, computer equipment and a machine readable medium. The method comprises the following steps: extracting a plurality of target features from a target image by using a neural network; associating a plurality of targets based on the plurality of extracted target features to form a structured target; and predicting classification confidence of the structured target and a regression value of the structured target, and displaying the structured target in the target image according to a prediction result. According to the invention, the subsequent tracking identification effect can be improved while the target association accuracy is improved. According to the method, multi-target independent detection tasks such as human body detection, human face detection and human head detection can be directly integrated into a structured target for detection, the correlation of the human body target, the human head target and the human face target in the same person is reserved, the error of target correlation in the tracking and recognition stage is reduced, and the tracking and recognition effect is improved.

Description

technical field [0001] The present invention relates to the technical field of image recognition, in particular to an object detection method, system, computer equipment and machine-readable medium. Background technique [0002] The structured detection of personnel includes human body detection, face detection, head detection, etc., which are of great significance in the application of face recognition, crowd counting, and behavior analysis and recognition in the field of monitoring and security. At present, the main way of structured detection of personnel is basically inclined to multi-task single detection based on deep learning. This method is easily affected by light, weather, occlusion, etc. , the variability of the attitude angle makes it easier to have a head without a human body or a detection frame without other detection frames, indicating that this method ignores the structural information of the person, that is, the correlation between the face, head and human ...

AI Technical Summary

Problems solved by technology

At present, the main method of structured detection of personnel tends to be multi-task individual detection based on deep learning. This method is easily affected by light, weather, occlusion, etc. , the variability of the attitude angle makes it easier to have a head without a human body or a detection frame without other detection frames, indicating that this method ignores the structural information of the person, that is, the correlation between the face, head and human body, resulting in To a certain extent, it affects the detection effect

Images