A Single Target Tracking Method Based on Siamese Network

Abstract

The invention discloses a single target tracking method based on a Siamese network, belonging to the technical field of target tracking. The present invention first constructs the neural network part of the Siamese network and trains the weight of the Siamese convolutional neural network. In the training process, the neural network model based on the embedded loss function is carried out. At the same time, the features of each layer are fused, and the stochastic gradient descent algorithm is used for loss. Optimization, and then obtain the results of classification and regression through RPN, and finally track the subsequent frames for the results of classification and regression. The invention can better detect and locate the tracking target, can effectively integrate the image detection method into the target tracking, and uses the image frame instead of the video, thereby reducing the training cost and calculation overhead. Thereby, the tracking processing efficiency is improved, and the effect of distinguishing similar objects is more obvious.

Description

technical field [0001] The invention belongs to the technical field of target tracking, in particular to a single target tracking technology based on Siamese network prediction technology. Background technique [0002] With the rapid development of hardware, software and artificial intelligence, target tracking has become one of the hot spots in the field of computer vision research and has been widely used. The tracking and focusing of cameras and the automatic target tracking of drones all require target tracking technology. In addition, there are specific object tracking, such as human body tracking, vehicle tracking in traffic monitoring systems, face tracking and gesture tracking in intelligent interaction systems. Simply put, target tracking is to establish the positional relationship of the object to be tracked in a continuous video sequence, and obtain the complete motion trajectory of the object. Given the coordinate position of the target in the first frame of th...

AI Technical Summary

Problems solved by technology

[0009] (1) Lack of necessary template updates;

[0010] (2) Since the window width remains unchanged during the tracking process, when the target scale changes, the tracking will fail;

[0011] (3) When the target speed is fast, the tracking effect is not good;

[0012] (4) The histogram feature is slightly deficient in the description of the target color feature, and lacks spatial information

Images