Search papers, labs, and topics across Lattice.
This paper investigates the issue of object-driven shortcuts in Zero-Shot Compositional Action Recognition (ZS-CAR), where models rely on object classes rather than temporal evidence to predict actions. The authors identify that sparse compositional supervision and the asymmetry in verb-object learning contribute to this problem, leading to poor generalization on novel combinations. They introduce Robust COmpositional REpresentations (RCORE), which incorporates Co-occurrence Prior Regularization and Temporal Order Regularization for Composition, resulting in significant improvements in compositional generalization across benchmark datasets Sth-com and EK100-com.
Object-driven shortcuts can severely undermine action recognition, but RCORE effectively mitigates this issue, enhancing model generalization to unseen verb-object combinations.
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.
