A Study on the Automation of Fish Species Recognition and Body Length Measurement System

Abstract

The rapid depletion of fishery resources has led to the global implementation of Total Allowable Catch (TAC) systems. However, the current manual survey methods employed by land-based inspectors show limitations in accuracy and efficiency. This study proposes an automated system for fish species recognition and body length measurement, utilizing the RT-DETR (Real-Time Detection Transformer) model and ARCore technology to address these issues. The proposed system employs smartphone Time of Flight (ToF) functionality to measure object distance and automatically calculates the weight of 11 TAC-managed fish species by measuring their body length and height. Experimental results reveal that the RT-DETR-x model outperformed the YOLOv8x model by achieving an average mAP50 value 2.3% higher, with a mean recognition accuracy of 96.5% across the 11 species. Furthermore, the ARCore-based length measurement technique exhibited over 95% accuracy for all species. This system is expected to minimize data omissions and streamline labor-intensive processes, thereby contributing to the efficient operation of the TAC system and sustainable management of fishery resources. The study presents an innovative approach that significantly enhances the accuracy and efficiency of fishery resource management, providing a crucial technological foundation for the advancement of future fisheries management policies.