Hierarchical Action Learning for Weakly-Supervised Action Segmentation

TL;DR

This paper introduces HAL, a hierarchical model that leverages the different evolution rates of visual and action variables to improve weakly-supervised action segmentation in videos.

Contribution

The paper proposes a novel hierarchical causal model with a pyramid transformer and alignment processes to better identify high-level actions over time.

Findings

HAL significantly outperforms existing methods on benchmark datasets.

The model effectively captures hierarchical action structures.

Latent action variables are proven to be strictly identifiable.

Abstract

Humans perceive actions through key transitions that structure actions across multiple abstraction levels, whereas machines, relying on visual features, tend to over-segment. This highlights the difficulty of enabling hierarchical reasoning in video understanding. Interestingly, we observe that lower-level visual and high-level action latent variables evolve at different rates, with low-level visual variables changing rapidly, while high-level action variables evolve more slowly, making them easier to identify. Building on this insight, we propose the Hierarchical Action Learning (\textbf{HAL}) model for weakly-supervised action segmentation. Our approach introduces a hierarchical causal data generation process, where high-level latent action governs the dynamics of low-level visual features. To model these varying timescales effectively, we introduce deterministic processes to align these latent variables over time. The \textbf{HAL} model employs a hierarchical pyramid transformer to capture both visual features and latent variables, and a sparse transition constraint is applied to enforce the slower dynamics of high-level action variables. This mechanism enhances the identification of these latent variables over time. Under mild assumptions, we prove that these latent action variables are strictly identifiable. Experimental results on several benchmarks show that the \textbf{HAL} model significantly outperforms existing methods for weakly-supervised action segmentation, confirming its practical effectiveness in real-world applications.