target tracking method based on a SiameFC framework and a PFP neural network

Abstract

The invention belongs to the technical field of computer vision, and particularly relates to a target tracking method based on a Siamese FC framework and a PFP neural network, which comprises the following steps: (1) based on the Siamese FC framework, processing a target area in a first frame of a video to obtain template characteristics; Inputting the template features into a PFP neural network to obtain template final features; (2) based on a Siamese FC framework, performing t-detection on the t- Processing the target area in one frame to obtain a search area feature; Inputting the featuresof the search area into a PFP neural network to obtain final features of the search area; (3) taking the final feature of the template as a convolution kernel, performing convolution on the final feature of the search area, and determining a central position of a tracking target in the t-th frame and a target area; And (4) repeating the step (2) and the step (3) until the video is finished, and completing the tracking of the central position of the tracking target and the target area. According to the invention, the context information and different receptive field information can be fused, sothat the target tracking precision is improved.

Description

technical field [0001] The invention belongs to the technical field of computer vision, and in particular relates to a target tracking method based on a SiameseFC framework and a PFP neural network. Background technique [0002] Object tracking has become one of the most active researches in computer vision technology due to its wide application in many fields such as behavior analysis, vehicle navigation, human-computer interaction, medical imaging, and video surveillance. Target tracking refers to the target position in the first frame of a given video, and target positioning for each subsequent frame. The central problem of object tracking is to keep track of objects that change over time. Although the target tracking algorithm has been developed rapidly under the continuous research of scholars at home and abroad in recent years, it still cannot achieve good results under the conditions of drastic changes in illumination, fast movement of the target, and partial occlusi...

AI Technical Summary

Problems solved by technology

The SiameseFC proposed in 2016 uses the fully convolutional twin neural network to obtain the feature maps of the template image and the search area, and directly uses the feature map of the template image as a filter to obtain the target in the search area. SiameseFC realizes end-to-end training. The features extracted by the network are more suitable for target tracking, and it also solves the problem of boundary effects, but it can only use the feature map output by a single-layer neural network, and does not integrate context information and information of different receptive fields well. Limitations limit its ability to distinguish objects from background and pinpoint objects

[0005] In view of the fact that the current tracking algorithm only uses the features output by the last layer of the neural network, and cannot well integrate the context information and the information of different receptive fields, it is necessary to design a tracking algorithm so that it can well integrate the context information and the information of different receptive fields. Information, so as to better predict the position of the target, so that the tracking accuracy is improved

Images