Diffusion-Based Action Recognition Generalizes to Untrained Domains

Rogério Guimarães, Frank Xiao, Pietro Perona, Markus Marks; Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2026, pp. 5919-5933

Humans can recognize the same actions despite large context and viewpoint variations, such as differences between species (walking in spiders vs. horses), viewpoints (egocentric vs. third-person), and contexts (real life vs movies). Current deep learning models struggle with such generalization. We propose using features generated by a Vision Diffusion Model (VDM), aggregated via a transformer, to achieve human-like action recognition across these challenging conditions. We find that generalization is enhanced by the use of a model conditioned on earlier timesteps of the diffusion process to highlight semantic information over pixel level details in the extracted features. We experimentally explore the generalization properties of our approach in classifying actions across animal species, across different viewing angles, and different recording contexts. Our model sets a new state-of-the-art across all three generalization benchmarks, bringing machine action recognition closer to human-like robustness.