YOLOv7 and EfficientNet Based Program For Detecting and Classifying Fish in Aquaculture Environment

The potential for aquaculture to supply the world’s demand for seafood is enormous. Effective fish monitoring and management in aquaculture, on the other hand, continues to be a serious challenge. This report proposed a study to develop a fish detection and classification system. We used YOLOv7 to detect fish and EfficientNet to classify the fish species. We trained and tested the models on augmented images of 9 different species of fish with 1000 images in each class. The proposed approach has a mAP score of 0.994 while the accuracy of the model was 0.9919 on the train set and 0.8139 on the test set. Our result shows that the combination of YOLOv7 and EfficientNet is applicable for efficient fish identification and categorization in the aquaculture environment.