WildFishNet Open Set Wild Fish Recognition Deep Neural Network with Fusion Activation Pattern

WildFishNet introduces a breakthrough in wild fish species recognition using the MobileNet deep learning model, addressing biodiversity monitoring and fisheries management challenges. This innovative system features a Fusion Activation Pattern (FAP) that combines multiple data sources to enhance classification accuracy and robustness, crucial for managing diverse habitats and environmental conditions. Unlike traditional methods limited by manual classification and species coverage, WildFishNet efficiently identifies and classifies over 29 wild fish species, including salmon, trout, and carp, and handles open-set recognition problems by distinguishing between known and novel species. This project has been rigorously tested to demonstrate its versatility across different aquatic environments, significantly advancing the capabilities of ecological monitoring and conservation efforts. Key sections of the study include related work, methodology, experimental results, and future research directions, positioning WildFishNet as a substantial contribution to the field of aquatic ecosystem management. 

Keywords: Wild fish recognition, MobileNet, deep learning, biodiversity monitoring, ecological conservation.