Action Segmentation Using 2D Skeleton Heatmaps and Multi-Modality Fusion

TL;DR

This paper introduces a novel 2D skeleton heatmap-based action segmentation method using TCNs, which outperforms some 3D-based methods and explores multi-modality fusion with RGB videos for fine-grained human activity recognition.

Contribution

It is the first to utilize 2D skeleton heatmaps and fuse them with RGB videos for action segmentation, demonstrating improved robustness and performance.

Findings

Comparable or superior performance to 3D skeleton methods

Enhanced robustness against missing keypoints

First to explore 2D skeleton+RGB fusion for this task

Abstract

This paper presents a 2D skeleton-based action segmentation method with applications in fine-grained human activity recognition. In contrast with state-of-the-art methods which directly take sequences of 3D skeleton coordinates as inputs and apply Graph Convolutional Networks (GCNs) for spatiotemporal feature learning, our main idea is to use sequences of 2D skeleton heatmaps as inputs and employ Temporal Convolutional Networks (TCNs) to extract spatiotemporal features. Despite lacking 3D information, our approach yields comparable/superior performances and better robustness against missing keypoints than previous methods on action segmentation datasets. Moreover, we improve the performances further by using both 2D skeleton heatmaps and RGB videos as inputs. To our best knowledge, this is the first work to utilize 2D skeleton heatmap inputs and the first work to explore 2D skeleton+RGB fusion for action segmentation.