MSR-YOLO: Method to Enhance Fish Detection and Tracking in Fish Farms
Hussam El-Din Mohamed, Ali Fadl, Omar Anas, Youssef Wageeh, Noha ElMasry, Ayman Nabil, Ayman Atia
Abstract
Tasks involving the monitoring of fish farms such as controlling fish ponds is one of the expensive and difficult tasks for fish farmers. Usually, fish farmers are doing these tasks manually which costs them time and money. We propose a system that automates the monitoring of the fish farm. This paper presents a technique to enhance the detection of fish and their trajectories in challenging water conditions. Firstly, we used image enhancement techniques to enhance unclear water images and to better identify fish. Then, we applied an object detection algorithm to detect fish. Finally, the detected objects’ coordinates are then used to extract features like count and trajectories. All experiments were done on our experimental setup. The technique showed promising results in regards to detection and tracking accuracy when applied.