<![CDATA[Accurate monitoring of mouse behavior in the Elevated Plus Maze (EPM) is essential for anxiety-related biomedical research, yet manual observation is time-consuming, subjective, and prone to human error. This study proposes an optimized automated tracking framework that integrates YOLOv8 detection with tracking methods including an adaptive Kalman Filter and DeepSORT, and compares them with conventional trackers such as CSRT and GOTURN. System performance was evaluated using Intersection over Union (IoU), Center Location Error (CLE), and Frames Per Second (FPS), with the Weighted Scoring Method (WSM) used for overall performance comparison. Experimental results show that the proposed YOLOv8 with adaptive Kalman filtering (frame interval = 5) provides the best balance between accuracy and computational efficiency. The approach achieved an IoU of 0.89 and CLE of 2.34 while increasing processing speed from 10.44 FPS to 22.55 FPS, representing an improvement of approximately 116% compared with the baseline configuration. Despite a slight increase in failure rates, the framework maintained stable real-time tracking performance under laboratory conditions. These results demonstrate that the proposed system improves both tracking efficiency and robustness, offering a reliable automated solution for high-throughput behavioral monitoring. The framework is particularly suitable for laboratory automation environments, supporting more objective behavioral assessment and improved data integrity in preclinical biomedical research.]]>
Copyrights © 2026
