Computer Vision-based Automated System for Animal Behavior Analysis

The TaiLOR system streamlines rodent behavior analysis by addressing limitations in manual tracking and existing video analysis tools. Leveraging methods like SAM2 masks and hierarchical rule-based tagging, TaiLOR achieves high accuracy in behavioral segmentation. Results demonstrate improved tagging for maternal care, social studies, and cognitive tasks like the Barnes Maze. Background analysis highlights the shortcomings of SimBA and DeepLabCut while showcasing the need for tailored...