Timestamp-Supervised Action Segmentation with Graph Convolutional Networks

TL;DR

This paper presents a novel timestamp-supervised action segmentation method using graph convolutional networks that iteratively refines dense frame labels from sparse annotations, outperforming baseline models and matching state-of-the-art results.

Contribution

Introduces an end-to-end trainable graph convolutional network framework for timestamp-supervised action segmentation with an alternating learning process.

Findings

Outperforms multi-layer perceptron baseline

Achieves comparable or better results than state-of-the-art methods

Validated on four public datasets with consistent improvements

Abstract

We introduce a novel approach for temporal activity segmentation with timestamp supervision. Our main contribution is a graph convolutional network, which is learned in an end-to-end manner to exploit both frame features and connections between neighboring frames to generate dense framewise labels from sparse timestamp labels. The generated dense framewise labels can then be used to train the segmentation model. In addition, we propose a framework for alternating learning of both the segmentation model and the graph convolutional model, which first initializes and then iteratively refines the learned models. Detailed experiments on four public datasets, including 50 Salads, GTEA, Breakfast, and Desktop Assembly, show that our method is superior to the multi-layer perceptron baseline, while performing on par with or better than the state of the art in temporal activity segmentation with timestamp supervision.