Mapping Rats Movements: Tracking and Visualizing Heatmaps and Trajectories Using Computer Vision

Abstract

Rats play a crucial role in pharmacological research, serving as valuable models for studying the effects of drugs and treatments on various physiological and behavioural parameters. The proposed system effectively tracks, detects, and maintains rats' identities while also extracting heatmaps and full movement trajectories for extensive analysis. Our approach leverages the YOLOv5 model for accurate rat detection and DeepSORT for reliable tracking. Each rat is assigned a unique ID, and the system includes a proposed tracking region to maintain identification despite occlusions or undetected frames. The tracking experiment achieved a high MOTA score of 99.24%, demonstrating the system's robustness and accuracy in diverse experimental conditions. Our methodology proves effective in maintaining precise identification even in challenging scenarios, highlighting its potential for enhancing pharmacological research.