Target tracking method based on spatial offset learning

Abstract

The invention provides a target tracking method based on spatial offset learning, which relates to the technical field of computer vision. The target object to be tracked is selected and determined, and the target selection process is automatically extracted or manually specified by the moving target detection method. Spatial migration learning network includes four parts: image data extraction, depth neural network, multi-layer perceptron MLP and spatial migration output. In real-time processing, the video image collected by camera and stored in storage area is extracted as the input image tobe tracked. In the case of offline processing, the captured video files are decomposed into image sequences composed of multiple frames. Short-time tracking adopts particle filter, in which a particle represents a possible target image block, and the corresponding target attention area ROI is used as the online training set of the spatial offset learning network, and the stochastic gradient descent (SGD) method is used to online train the spatial offset learning network and update the network parameters. Target Location and updating are carried out.

Description

technical field [0001] The invention relates to the technical fields of computer vision, graphic images, pattern recognition and machine learning. Background technique [0002] Visual object tracking is an important research topic in the field of computer vision. Its main task is to obtain the continuous position, appearance and motion information of the object, and then provide the basis for further semantic analysis (such as behavior recognition, scene understanding, etc.). Target tracking research is widely used in intelligent monitoring, human-computer interaction, automatic control systems and other fields, and has strong practical value. At present, target tracking methods mainly include classical target tracking methods and deep learning target tracking methods. [0003] The classic target tracking methods are mainly divided into two categories: Generative Methods and Discriminative Methods. The generative method assumes that the target can be expressed through some...

AI Technical Summary

Problems solved by technology

The research on deep learning object tracking is also developing very rapidly, but due to the lack of prior knowledge and real-time requirements in object tracking, it is difficult to fully display the deep learning technology based on a large amount of training data and parameter calculation. a lot of room to explore

Images