Action Segmentation with Mixed Temporal Domain Adaptation

TL;DR

This paper introduces Mixed Temporal Domain Adaptation (MTDA), a novel method for action segmentation that aligns features across domains at both frame and video levels, improving performance on challenging datasets.

Contribution

The paper proposes MTDA, the first approach to jointly adapt frame- and video-level features in action segmentation, incorporating a domain attention mechanism for better alignment.

Findings

MTDA outperforms state-of-the-art methods on GTEA, 50Salads, and Breakfast datasets.

Achieves 6.4% gain on F1@50 and 6.8% gain on edit score for GTEA.

Effective domain adaptation with large margins across multiple datasets.

Abstract

The main progress for action segmentation comes from densely-annotated data for fully-supervised learning. Since manual annotation for frame-level actions is time-consuming and challenging, we propose to exploit auxiliary unlabeled videos, which are much easier to obtain, by shaping this problem as a domain adaptation (DA) problem. Although various DA techniques have been proposed in recent years, most of them have been developed only for the spatial direction. Therefore, we propose Mixed Temporal Domain Adaptation (MTDA) to jointly align frame- and video-level embedded feature spaces across domains, and further integrate with the domain attention mechanism to focus on aligning the frame-level features with higher domain discrepancy, leading to more effective domain adaptation. Finally, we evaluate our proposed methods on three challenging datasets (GTEA, 50Salads, and Breakfast), and validate that MTDA outperforms the current state-of-the-art methods on all three datasets by large margins (e.g. 6.4% gain on F1@50 and 6.8% gain on the edit score for GTEA).