Contrastive Learning with Cross-Modal Knowledge Mining for Multimodal Human Activity Recognition

TL;DR

This paper introduces a novel multimodal self-supervised learning framework called CMC-CMKM for human activity recognition, leveraging cross-modal knowledge to improve recognition accuracy across various scenarios.

Contribution

The paper proposes a flexible framework that enhances contrastive self-supervised learning by incorporating cross-modal knowledge mining for multimodal human activity recognition.

Findings

Significantly outperforms unimodal and multimodal baselines in experiments.

Achieves competitive performance with supervised methods.

Effective in fully-supervised, retrieval, and semi-supervised scenarios.

Abstract

Human Activity Recognition is a field of research where input data can take many forms. Each of the possible input modalities describes human behaviour in a different way, and each has its own strengths and weaknesses. We explore the hypothesis that leveraging multiple modalities can lead to better recognition. Since manual annotation of input data is expensive and time-consuming, the emphasis is made on self-supervised methods which can learn useful feature representations without any ground truth labels. We extend a number of recent contrastive self-supervised approaches for the task of Human Activity Recognition, leveraging inertial and skeleton data. Furthermore, we propose a flexible, general-purpose framework for performing multimodal self-supervised learning, named Contrastive Multiview Coding with Cross-Modal Knowledge Mining (CMC-CMKM). This framework exploits modality-specific knowledge in order to mitigate the limitations of typical self-supervised frameworks. The extensive experiments on two widely-used datasets demonstrate that the suggested framework significantly outperforms contrastive unimodal and multimodal baselines on different scenarios, including fully-supervised fine-tuning, activity retrieval and semi-supervised learning. Furthermore, it shows performance competitive even compared to supervised methods.