Multi-modal AI Systems for Human and Animal Pose Estimation in Challenging Conditions

Abstract

This paper explores the development of multi-modal AI systems for pose estimation in challenging conditions for both humans and animals. Existing single-modality approaches struggle in challenging scenarios such as emergency response and wildlife observation due to factors like smoke, low light, obstacles, and long-distance observations. To address these challenges, this research proposes integrating multiple sensor modalities and leveraging the strengths of different sensors to enhance accuracy and robustness in pose estimation.