IMG2IMU: Translating Knowledge from Large-Scale Images to IMU Sensing Applications

TL;DR

This paper introduces IMG2IMU, a method that leverages large-scale image pre-training by converting IMU data into spectrograms, enabling improved performance on diverse IMU sensing tasks with limited data.

Contribution

The paper proposes a novel approach to transfer vision pre-training to IMU sensing by converting sensor data into spectrograms and introducing a sensor-aware contrastive pre-training method.

Findings

Outperforms sensor-only pre-trained baselines by 9.6% F1-score on average

Effective transfer of vision knowledge to IMU tasks with limited data

Demonstrates versatility across four different IMU sensing applications

Abstract

Pre-training representations acquired via self-supervised learning could achieve high accuracy on even tasks with small training data. Unlike in vision and natural language processing domains, pre-training for IMU-based applications is challenging, as there are few public datasets with sufficient size and diversity to learn generalizable representations. To overcome this problem, we propose IMG2IMU that adapts pre-trained representation from large-scale images to diverse IMU sensing tasks. We convert the sensor data into visually interpretable spectrograms for the model to utilize the knowledge gained from vision. We further present a sensor-aware pre-training method for images that enables models to acquire particularly impactful knowledge for IMU sensing applications. This involves using contrastive learning on our augmentation set customized for the properties of sensor data. Our evaluation with four different IMU sensing tasks shows that IMG2IMU outperforms the baselines pre-trained on sensor data by an average of 9.6%p F1-score, illustrating that vision knowledge can be usefully incorporated into IMU sensing applications where only limited training data is available.