PROTOTIPE ALAT PENYORTIR JENIS IKAN OTOMATIS BERDASARKAN CITRA MATA MENGGUNAKAN ALGORITMA YOLO

Salsabil, Sani
Anjara, Febria
Jaenul, Ariep
Dionova, Brainvendra Widi
Putri, Arisa Olivia
Salsabila, Ghina

JTERA (Jurnal Teknologi Rekayasa)
Vol 9, No 2: December 2024

This research presents the development of an automated fish sorting system leveraging image processing technology to enhance productivity within the fishing industry. Conventional fish sorting methods are predominantly manual, resulting in significant time and labor expenditures. To mitigate these challenges, the researchers designed and implemented a fish sorting device capable of categorizing fish based on species and quality during the management of fishery products. The system employs parameters such as eye color and skin or scale color, utilizing the YOLO algorithm to evaluate the accuracy of species and quality detection through image processing. The research encompasses several testing scenarios to validate the system's efficacy. The initial test utilized the ESP32 Cam to identify fish types from a sample of 16 fish, including Milkfish, Mackerel, Catfish, and Tilapia. The results demonstrated a high level of accuracy, with detection readings ranging from 0.902344 to 0.996094; however, Catfish and Tilapia were not detected due to their exclusion from the training dataset. Following this, the study evaluated fish quality through scale color parameters using the TCS3200 sensor, indicating that RGB values effectively differentiate fresh and non-fresh fish. The final test assessed actuator performance for sorting, achieving a 100% accuracy rate with no operational errors in the servo motor. Overall, the results indicate that the system can accurately identify species with a 100% success rate across the 16 samples and achieve the same rate for quality identification from 8 test images of two fish species. Additionally, the AI training metrics yielded a precision of 0.98 (98%), a recall of 0.96 (96%), and an F1 score of 0.97 (97%). In conclusion, this automated fish sorting system demonstrates significant potential for improving efficiency and accuracy in fish classification and quality assessment within the fishing industry, thereby contributing to enhanced productivity and operational effectiveness.