Improved YOLOv8 Underwater Object Detection Combined with CBAM

Marine organism detection plays a crucial role in the development of marine resources and the maintenance of global ecological balance. In complex underwater environments, traditional object detection algorithms often perform inadequately. This paper employs deep learning object detection algorithms that leverage big data by integrating a Convolutional Block Attention Module (CBAM) into the detection head of YOLOv8, along with a new loss function, for the detection of underwater organisms. Our goal is to enhance the model's focus on the key features of marine organisms, thereby improving detection performance and accuracy. We utilized marine image data from publicly available datasets for training and testing. Experimental results demonstrate that the improved YOLOv8 achieves a 2.3% increase in mAP50 compared to the original YOLOv8, with a slight improvement in the recognition accuracy of various organisms, thereby validating the effectiveness of the proposed algorithm. This provides an effective solution for the field of marine organism recognition.