Receptive Field Enhancement and Attention Feature Fusion Network for Underwater Object Detection

Abstract Underwater environments have characteristics such as unclear imaging and complex backgrounds, which lead to poor performance when applying mainstream object detection models directly. To improve the accuracy of underwater object detection, we propose a novel object detection model RF-YOLO, which uses Receptive Field Enhancement Module(RFAM)in the backbone network to finish receptive field enhancement and extract more effective features. We design Free-channel iterative Attention Feature Fusion༈FAFF༉ module to reconstruct the neck network and fuse different scales of feature layers to achieve cross-channel attention feature fusion. We use SIoU as the loss function of the model, which makes the model converge to the optimal direction of training through angle cost, distance cost, shape cost, and IoU cost. The network parameters increase after adding modules, and the model is not easy to converge to the optimal state, so we propose a new training method, which effectively mines the performance of the detection network. Experiments show that the proposed RF-YOLO achieves mAP of 87.56% and 86.39% on URPC2019 and URPC2020 respectively. Through comparative experiments and ablation experiments, it was verified that the proposed network model has higher detection accuracy in complex underwater environments.