A Mobile App for Counting Shrimp Larvae Based on the YOLO V5 Method

Abstract

Manual counting of shrimp larvae in aquaculture is labour-intensive and time-consuming. This study aims to develop a mobile application to automate the counting process using an object detection algorithm. The application features dual functionality for real-time camera capture and image upload. Model performance was evaluated using several metrics, including Mean Average Precision, precision, and recall. The object detection model achieved a Mean Average Precision (mAP) of 93.93%, precision of 91%, and recall of 89.3%. Trials of the application demonstrated an average accuracy rate of 91.03% in detecting shrimp larvae. Despite challenges in detecting transparent larvae and distinguishing them from debris, the results indicate that the application holds promise for enhancing efficiency in shrimp farming operations. Future improvements may be directed towards enhancing application performance by refining the dataset and tuning model parameters to increase recall without compromising precision. This study represents a significant step towards integrating AI-driven technologies into aquaculture, potentially transforming the shrimp larvae counting and management process in the industry.