Interpretable Motion-Attentive Maps: Spatio-Temporally Localizing Concepts in Video Diffusion Transformers

TL;DR

This paper introduces a novel method for localizing motion concepts in videos generated by Diffusion Transformers, providing interpretable saliency maps without requiring gradient updates, and demonstrating superior localization and segmentation performance.

Contribution

It proposes GramCol for spatial localization and a motion-feature selection algorithm to create Interpretable Motion-Attentive Maps, advancing interpretability in video diffusion models.

Findings

Effective motion localization in videos

Zero-shot video semantic segmentation capability

Clearer, interpretable saliency maps for motion and non-motion concepts

Abstract

Video Diffusion Transformers (DiTs) have been synthesizing high-quality video with high fidelity from given text descriptions involving motion. However, understanding how Video DiTs convert motion words into video remains insufficient. Furthermore, while prior studies on interpretable saliency maps primarily target objects, motion-related behavior in Video DiTs remains largely unexplored. In this paper, we investigate concrete motion features that specify when and which object moves for a given motion concept. First, to spatially localize, we introduce GramCol, which adaptively produces per-frame saliency maps for any text concept, including both motion and non-motion. Second, we propose a motion-feature selection algorithm to obtain an Interpretable Motion-Attentive Map (IMAP) that localizes motion spatially and temporally. Our method discovers concept saliency maps without the need for any gradient calculation or parameter update. Experimentally, our method shows outstanding localization capability on the motion localization task and zero-shot video semantic segmentation, providing interpretable and clearer saliency maps for both motion and non-motion concepts.