Uncertainty-aware Score Distribution Learning for Action Quality Assessment

Assessing action quality from videos has attracted growing attention in recent years. Most existing approaches usually tackle this problem based on regression algorithms, which ignore the intrinsic ambiguity in the score labels caused by multiple judges or their subjective appraisals. To address this issue, we propose an uncertainty-aware score distribution learning (USDL) approach for action quality assessment (AQA). Specifically, we regard an action as an instance associated with a score distribution, which describes the probability of different evaluated scores. Moreover, under the circumstance where fine-grained score labels are available (e.g., difficulty degree of an action or multiple scores from different judges), we further devise a multi-path uncertainty-aware score distributions learning (MUSDL) method to explore the disentangled components of a score. We conduct experiments on three AQA datasets containing various Olympic actions and surgical activities, where our approaches set new state-of-the-arts under the Spearman's Rank Correlation.

PDF Abstract CVPR 2020 PDF CVPR 2020 Abstract

Results from the Paper

Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Action Quality Assessment AQA-7 USDL Spearman Correlation 81.02% # 4
RL2(*100) 2.57 # 2
Action Quality Assessment AQA-7 I3D+MLP Spearman Correlation 74.72% # 6
Action Quality Assessment MTL-AQA USDL Spearman Correlation 90.66 # 12
RL2(*100) 0.609 # 7
Action Quality Assessment MTL-AQA MUSDL Spearman Correlation 91.58 # 11
RL2(*100) 0.654 # 8
Action Quality Assessment MTL-AQA MUSDL(w/ DD) Spearman Correlation 92.73 # 7
RL2(*100) 0.451 # 4
Action Quality Assessment MTL-AQA USDL(w/ DD) Spearman Correlation 92.31 # 8
RL2(*100) 0.468 # 6
Action Quality Assessment MTL-AQA I3D+MLP Spearman Correlation 89.21 # 15


No methods listed for this paper. Add relevant methods here