Action Recognition with Domain Invariant Features of Skeleton Image

TL;DR

This paper introduces a CNN-based action recognition method that employs domain adversarial training to learn view-invariant features from skeleton images, enhancing generalization across different subjects and viewpoints.

Contribution

It proposes a novel two-level domain adversarial learning framework to improve skeleton image-based action recognition by reducing view and subject dependency.

Findings

Achieves 2.4% and 1.9% accuracy improvements over baseline in cross-subject and cross-view tests.

Demonstrates competitive performance on NTU RGB+D dataset.

Enhances generalization of skeleton-based action recognition models.

Abstract

Due to the fast processing-speed and robustness it can achieve, skeleton-based action recognition has recently received the attention of the computer vision community. The recent Convolutional Neural Network (CNN)-based methods have shown commendable performance in learning spatio-temporal representations for skeleton sequence, which use skeleton image as input to a CNN. Since the CNN-based methods mainly encoding the temporal and skeleton joints simply as rows and columns, respectively, the latent correlation related to all joints may be lost caused by the 2D convolution. To solve this problem, we propose a novel CNN-based method with adversarial training for action recognition. We introduce a two-level domain adversarial learning to align the features of skeleton images from different view angles or subjects, respectively, thus further improve the generalization. We evaluated our proposed method on NTU RGB+D. It achieves competitive results compared with state-of-the-art methods and 2.4$\%$, 1.9$\%$ accuracy gain than the baseline for cross-subject and cross-view.