Multi-instance learning idea-based training sample selection method during target tracking process

Abstract

The invention provides a multi-instance learning idea-based training sample selection method during the target tracking process and relates to the target tracking technology in the field of computer vision. According to the technical scheme of the invention, the problem that poorly marked samples cannot be accurately marked during the target tracking process due to the conventional supervised learning algorithm in the prior art can be solved. According to the method, all positive samples are put into a positive sample packet. Samples having a smaller contribute to a log likelihood function with respect to the packet are considered to be poor samples. In the iteration mode, a worst sample is removed out of the packet during each iteration process till a sufficient number of samples are remained in the positive sample packet. Therefore, the method is superior to other assessment algorithms in the aspect of tracking accuracy. Meanwhile, the method is higher in frame rate and satisfies real-time requirements.

Description

technical field [0001] The invention relates to target tracking technology in the field of computer vision. Background technique [0002] Although the tracking algorithm for specific objects such as faces and pedestrians has achieved a series of successes, due to the lack of sufficient prior information for an arbitrary object, and the appearance of the object will change due to deformation, rotation or illumination changes during the tracking process. Due to drastic changes and other reasons, the target tracking algorithm for general objects still faces many challenges. Because the tracking algorithm for general targets is not reliable enough, the development of many application scenarios based on general tracking is limited. For example, the target tracking function in the camera module of a smart phone often requires the target to have relatively obvious features, or can only achieve short-term tracking of the target. [0003] Although more and more researchers believe ...

AI Technical Summary

Problems solved by technology

[0007] The present invention aims to solve the defect of "weakly labeled" samples faced by traditional supervised learning algorithms in target tracking, resulting in the inability to guarantee accurate labeling of samples, thereby providing a training sample selection method based on the idea of ​​multi-instance learning in target tracking

Images