Multi-target number selection tracking method

Abstract

The invention relates to a real-time autonomous tracking method based on deep learning. a computer vision target detection and computer vision target tracking algorithm based on an artificial neural network of deep learning is provided; A high-performance computing unit can be utilized to run an arithmetic unit of a neural network to detect targets, then a target tracking algorithm is run to trackall the targets at the same time, and a specific target can be manually intervened and selected to focus on single-person tracking. Compared with a traditional single-target tracking algorithm, traditional single-target tracking needs to manually select a target in a framing manner, but for a moving target, framing failure is often caused by operation delay. According to the algorithm, inaccurateframe selection and target deviation caused by operation delay of manual frame selection of the target are avoided. A camera-server architecture is constructed, all target data in the camera are processed at the same time, multi-person tracking and single-person continuous tracking in the whole area are achieved, and experimental results show that real-time neural network operation can be achieved, and then the two-step tracking effect of detection and person selection is achieved in combination with a target tracking algorithm.

Description

technical field [0001] The invention belongs to the technical field of computer vision, and in particular relates to a method for selecting and tracking multiple target numbers. Background technique [0002] With the promotion of Safe City and the strategy of strengthening the police with science and technology, the construction of video surveillance system is gradually developing towards scale, network, intelligence and actual combat. At this stage, the video surveillance network is spreading rapidly, but with the expansion of the scale of the video surveillance system, the massive data it possesses also brings difficulties in processing. When it is necessary to track a specific target, it is often only possible to rely on manual observation of the monitoring screen. [0003] Object detection is an important research content of machine vision. The traditional target detection process first locates the target position on the input image, then extracts features from the tar...

AI Technical Summary

Problems solved by technology

However, with the increase of its accuracy rate, the amount of calculation has increased explosively, which has reduced the running speed of the target tracking algorithm to a few seconds per frame in recent years, and it cannot be put into practical use at all.

Images