Optical Mouse: 3D Mouse Pose From Single-View Video

Many human clinical conditions result in abnormal motion. When investigating corresponding animal models, researchers have to measure the motion as a 3D pose. Current approaches are expensive and cause high stress for the animals.

A recent paper presents a low-cost, non-invasive, computer-vision-based approach for continuously measuring the motion of laboratory mice.

The labeled 2D keypoints. Image credit: Hu, B., et al., arXiv:2106.09251

Firstly, 2D keypoints are predicted. Then, they are optimized for the 3D pose subject to priors learned from data. The lack of data for mice is successfully overcome by collecting new data and adapting human pose data where needed. It is shown that 3D joint angles enable predicting health-related attributes more easily than other features. The suggested system can replace custom hardware solutions in determining gait parameters such as stride length.

We present a method to infer the 3D pose of mice, including the limbs and feet, from monocular videos. Many human clinical conditions and their corresponding animal models result in abnormal motion, and accurately measuring 3D motion at scale offers insights into health. The 3D poses improve classification of health-related attributes over 2D representations. The inferred poses are accurate enough to estimate stride length even when the feet are mostly occluded. This method could be applied as part of a continuous monitoring system to non-invasively measure animal health.

Research paper: Hu, B., “Optical Mouse: 3D Mouse Pose From Single-View Video”, 2021. Link: https://arxiv.org/abs/2106.09251


Leave a Reply

Your email address will not be published. Required fields are marked *