Diffusion Action Segmentation

The University of Sydney, Peking University, University of Central Florida
ICCV 2023
MY ALT TEXT

Iterative Denoising on Action Sequences

Abstract

Temporal action segmentation is crucial for understanding long-form videos. Previous works on this task commonly adopt an iterative refinement paradigm by using multi-stage models. We propose a novel framework via denoising diffusion models, which nonetheless shares the same inherent spirit of such iterative refinement. In this framework, action predictions are iteratively generated from random noise with input video features as conditions. To enhance the modeling of three striking characteristics of human actions, including the position prior, the boundary ambiguity, and the relational dependency, we devise a unified masking strategy for the conditioning inputs in our framework. Extensive experiments on three benchmark datasets, i.e., GTEA, 50Salads, and Breakfast, are performed and the proposed method achieves superior or comparable results to state-of-the-art methods, showing the effectiveness of a generative approach for action segmentation.

MY ALT TEXT

Comparison with state-of-the-art methods. Our method achieves better results on 50Salads and Breakfast, and comparable performance on GTEA.

Visualization on GTEA

Visualization on 50Salads

Visualization on Breakfast

BibTeX


@inproceedings{liu2023diffusion,
  title={Diffusion Action Segmentation},
  author={Liu, Daochang and Li, Qiyue and Dinh, Anh-Dung and Jiang, Tingting and Shah, Mubarak and Xu, Chang},
  booktitle={International Conference on Computer Vision (ICCV)},
  year={2023}
}