Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces a lightweight dual-loss training framework for temporal action segmentation (TAS) that combines boundary regression and segment-level regularization. The boundary-regression loss uses a single-channel boundary prediction to improve temporal localization, while the CDF-based segment-level regularization loss encourages coherent within-segment structure. Experiments on three benchmark datasets demonstrate that this approach improves segment-level consistency and boundary quality, leading to higher F1 and Edit scores across multiple TAS models without significantly impacting frame-wise accuracy.
Achieve state-of-the-art fine-grained action segmentation with a surprisingly simple dual-loss framework that requires minimal architectural changes.
Recent progress in Temporal Action Segmentation (TAS) has increasingly relied on complex architectures, which can hinder practical deployment. We present a lightweight dual-loss training framework that improves fine-grained segmentation quality with only one additional output channel and two auxiliary loss terms, requiring minimal architectural modification. Our approach combines a boundary-regression loss that promotes accurate temporal localization via a single-channel boundary prediction and a CDF-based segment-level regularization loss that encourages coherent within-segment structure by matching cumulative distributions over predicted and ground-truth segments. The framework is architecture-agnostic and can be integrated into existing TAS models (e.g., MS-TCN, C2F-TCN, FACT) as a training-time loss function. Across three benchmark datasets, the proposed method improves segment-level consistency and boundary quality, yielding higher F1 and Edit scores across three different models. Frame-wise accuracy remains largely unchanged, highlighting that precise segmentation can be achieved through simple loss design rather than heavier architectures or inference-time refinements.