Intelligent Control for Automated Fish Feeding System in an Aquaponic Environment

Fish feeding in aquaponics presents challenges in accurately estimating and distributing feed, risking the health of both fish and plants due to potential water contamination. To address these issues, our study employs computer vision and deep learning to automate feed dispensing in aquaponics farms, ensuring precise feed quantities. We utilize the Mask R-CNN algorithm to identify fish measurements in tanks, aiding in feed quantity determination and minimizing over or underfeeding. By calculating a fish-to-feed ratio, we regulate feed amounts to enhance system performance. The remaining fish segment’s length and width are converted into weight, guiding feed calculations for each session. We conducted experiments with 432 Nile Tilapia, using their measurements to train the Mask R-CNN model. Results show a significant mean error of only 0.98% and 4.56% for length and weight predictions, respectively, with a high correlation coefficient of 0.97.