Precision Tracking and Enumeration of Benthic Species with Yolo+ Deepsort Network Improvements
Authors:-Dhanalakshmi V, Amirthajaya T, Bhuvaneshwari A
Abstract-The health of marine ecosystems is vital for biodiversity and ecological balance, with benthic species serving as key indicators of environmental conditions. Traditional monitoring methods, such as manual surveys and video analysis, are labor-intensive and error-prone. This research proposes an AI-driven solution integrating an optimized YOLO object detection model with an improved DeepSORT tracking algorithm for accurate benthic species identification and counting. The approach addresses challenges like poor visibility, occlusion, and species overlap, enhancing monitoring efficiency for conservation and fisheries management. Experimental results demonstrate an accuracy of 87.1% mAP@0.5 and 53.3% mAP@0.5:0.95, showing improvements of 1.8% and 4.0%, respectively, over YOLOv5. The integration of DeepSORT further strengthens its application in marine ranching supervision. This AI-based method offers a reliable and automated alternative for marine ecosystem evaluation and conservation efforts./p>
DOI: 10.61137/ijsret.vol.11.issue2.201
