FineCausal: A Causal-Based Framework for Interpretable Fine-Grained Action Quality Assessment

2025-03-31

AI Summary

Existing action quality assessment (AQA) models are largely black-box, suffering from background confounding and spurious inter-phase correlations, resulting in limited interpretability and robustness. To address these issues, this paper proposes the first causal evaluation framework for fine-grained AQA. It introduces a graph attention-based causal intervention module to explicitly disentangle human pose dynamics from background confounders, and a temporal causal attention module to model genuine causal dependencies among action phases. The method integrates causal intervention, graph attention networks (GAT), and fine-grained spatiotemporal representation learning. Evaluated on the FineDiving-HM dataset, it achieves state-of-the-art scoring accuracy while generating attributable visual feedback. This significantly enhances transparency, credibility, and interpretability of AQA—enabling principled, causally grounded quality assessment with diagnostic insights into both pose execution and contextual influences.

Technology Category

• Knowledge Representation and Reasoning: Action, Change, and Causality
• Reasoning under Uncertainty: Causality
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Action quality assessment (AQA) is critical for evaluating athletic performance, informing training strategies, and ensuring safety in competitive sports. However, existing deep learning approaches often operate as black boxes and are vulnerable to spurious correlations, limiting both their reliability and interpretability. In this paper, we introduce FineCausal, a novel causal-based framework that achieves state-of-the-art performance on the FineDiving-HM dataset. Our approach leverages a Graph Attention Network-based causal intervention module to disentangle human-centric foreground cues from background confounders, and incorporates a temporal causal attention module to capture fine-grained temporal dependencies across action stages. This dual-module strategy enables FineCausal to generate detailed spatio-temporal representations that not only achieve state-of-the-art scoring performance but also provide transparent, interpretable feedback on which features drive the assessment. Despite its strong performance, FineCausal requires extensive expert knowledge to define causal structures and depends on high-quality annotations, challenges that we discuss and address as future research directions. Code is available at https://github.com/Harrison21/FineCausal.

Problem

• Enhances interpretability in fine-grained action quality assessment
• Reduces reliance on spurious correlations in deep learning models
• Improves temporal and spatial feature representation for scoring

Innovation

• Graph Attention Network for causal intervention
• Temporal causal attention for dependencies
• Dual-module spatio-temporal representation generation