Play Fair: Frame Attributions in Video Models
This work provides a method for researchers and practitioners to understand which frames contribute to an action recognition model's prediction, which is an incremental improvement for explainable AI in video understanding.
This paper introduces a method to explain action recognition models by attributing a model's class score to individual frames in a video. They adapt the Shapley value for variable-length sequences, calling it Element Shapley Value (ESV), and propose a tractable approximation that scales linearly with the number of frames.
In this paper, we introduce an attribution method for explaining action recognition models. Such models fuse information from multiple frames within a video, through score aggregation or relational reasoning. We break down a model's class score into the sum of contributions from each frame, fairly. Our method adapts an axiomatic solution to fair reward distribution in cooperative games, known as the Shapley value, for elements in a variable-length sequence, which we call the Element Shapley Value (ESV). Critically, we propose a tractable approximation of ESV that scales linearly with the number of frames in the sequence. We employ ESV to explain two action recognition models (TRN and TSN) on the fine-grained dataset Something-Something. We offer detailed analysis of supporting/distracting frames, and the relationships of ESVs to the frame's position, class prediction, and sequence length. We compare ESV to naive baselines and two commonly used feature attribution methods: Grad-CAM and Integrated-Gradients.