Enhancing Warehouse Management Safety through YOLOv8 and BoT-SORT-Based Boundary Violation Detection Systems

• Published in: 2024 5th International Conference on Computer Engineering and Application (ICCEA) pp. 1568 - 1571
• Main Authors: Han, Xutian, Wang, Jianxiang, Sun, Zhihai
• Format: Conference Proceeding
• Language: English
• Published: IEEE 12.04.2024
• Subjects: Accuracy; BoT-SORT; Boundary Violation Identification; MOT; Pedestrians; Real-time systems; Safety; Surveillance; Target tracking; YOLO; YOLOv8
• ISSN: 2159-1288
• DOI: 10.1109/ICCEA62105.2024.10604198

In the contemporary landscape of warehouse management, ensuring the safety and integrity of stored goods and personnel is paramount. The integration of advanced computer vision technologies has opened new avenues for enhancing security measures within these environments. This study presents a novel approach to boundary violation detection, leveraging the robust capabilities of YOLOv8 and BoT-SORT algorithms. YOLOv8, the latest iteration in the line of 'You Only Look Once' object detection models, offers unprecedented accuracy and speed in identifying objects within complex scenes. Coupled with BoT-SORT, an algorithm specialized in multi-object tracking that incorporates both motion and appearance information, this system provides a comprehensive solution for real-time surveillance and security in warehouse settings.The results from the MOT20 dataset indicate that the multi-object tracking's precision has achieved a mark of 69.36%. Furthermore, it has been enhanced to 79.11% in addressing the challenges associated with object occlusions and the frequent switches in identities, showcasing significant effectiveness.It excels in tracking non-rigid pedestrian targets, offering superior accuracy compared to both YOLOv8-DeepSORT and YOLOv8-ByteTrack systems.