YOLOv7-Based Object Detection Model for Effective Aquarium

Aquariums provide controlled environments for studying aquatic ecosystems and diverse marine life under controlled conditions. However, effectively monitoring and comprehending these systems, particularly in monitoring species, real-time processing, scalability, resource efficiency, and conducting accurate studies, remains a significant challenge. YOLOv7, a state-of-the-art object detection algorithm, facilitates real-time object detection and counting, making it ideal for monitoring endangered species in aquariums. This paper proposes a YOLOv7-based object detection model for effective aquarium management. The suggested model solves problems by utilizing YOLOv7's strengths for rapid species detection, real-time processing, resource efficiency, robust model design, and domain adaptations. The PyTorch framework and V100 GPU-powered model achieve 87.56% precision on a marine life dataset, highlighting its classification accuracy. Its 85.46% recall rate shows its capacity to detect events. These measures yield an exceptional mean average precision (mAP) of 58.26% spanning a 0.5 to 0.95 confidence threshold range and 89.75% at 0.5. Using YOLOv7's capabilities, the suggested object identification technique addresses major aquarium management and research concerns, improving our understanding and protection of aquatic environments.