Why Can't I Open My Drawer? Mitigating Object-Driven Shortcuts in Zero-Shot Compositional Action Recognition

TL;DR

This paper identifies and mitigates object-driven shortcut learning in zero-shot compositional action recognition, proposing methods to improve models' reliance on temporal cues for better generalization to unseen verb-object combinations.

Contribution

It introduces RCORE, a framework with CPR and TORC components, to reduce shortcut learning and enhance compositional generalization in zero-shot action recognition.

Findings

RCORE reduces shortcut diagnostics in models.

RCORE improves generalization to unseen verb-object pairs.

Temporal cues are crucial for compositional action recognition.

Abstract

Zero-Shot Compositional Action Recognition (ZS-CAR) requires recognizing novel verb-object combinations composed of previously observed primitives. In this work, we tackle a key failure mode: models predict verbs via object-driven shortcuts (i.e., relying on the labeled object class) rather than temporal evidence. We argue that sparse compositional supervision and verb-object learning asymmetry can promote object-driven shortcut learning. Our analysis with proposed diagnostic metrics shows that existing methods overfit to training co-occurrence patterns and underuse temporal verb cues, resulting in weak generalization to unseen compositions. To address object-driven shortcuts, we propose Robust COmpositional REpresentations (RCORE) with two components. Co-occurrence Prior Regularization (CPR) adds explicit supervision for unseen compositions and regularizes the model against frequent co-occurrence priors by treating them as hard negatives. Temporal Order Regularization for Composition (TORC) enforces temporal-order sensitivity to learn temporally grounded verb representations. Across Sth-com and EK100-com, RCORE reduces shortcut diagnostics and consequently improves compositional generalization.