Action quality evaluation method based on uncertainty score distribution learning

Abstract

The invention discloses an action quality evaluation method based on uncertainty score distribution learning, and the method comprises the steps of taking a score label as a mean value, and generatingGaussian distribution as a supervision signal; sending the action video into a 3D neural network to predict a score label; optimizing the network by optimizing the KL divergence between the prediction score label and the supervision signal; and inputting the test video into the optimized prediction video score model, and selecting the score with the maximum probability value as the final prediction score. According to the method, the probability of the action quality score can be better described, and the ambiguity problem in the action quality evaluation score label is solved.

Description

technical field [0001] The invention relates to the technical fields of computer vision and deep learning, in particular to an action quality evaluation method based on uncertainty score distribution learning. Background technique [0002] Human action quality evaluation aims to evaluate the quality of a person's completion of a specific action. Due to its potential applications in sports video processing, medical care, etc., it has become an emerging and interesting topic in the field of computer vision. Action evaluation is more challenging than traditional action recognition problems that focus on classification because it needs to deal with videos with extremely small distance between classes. [0003] In the past few years, some methods for action quality evaluation have been proposed. However, for any action in the related art, the action quality evaluation is only regarded as a simple regression problem to predict the score of the video. Since the videos that need ...

AI Technical Summary

Problems solved by technology

[0006] However, this calculation method accounts for the ambiguity of the final score due to different ratings

Furthermore, the subjective evaluation of each judge may also bring uncertainty to the final score

In addition to diving sports, this phenomenon also widely exists in the evaluation of other sports (such as gymnastics, alpine skiing, etc.), making robust action quality evaluation particularly difficult

Images