A Target Detection Algorithm for Auto Drive Based on SRF-YOLOv7

The SRF-YOLOv7 algorithm based on YOLOv7 network model is proposed to solve the problem of false detection and missing detection caused by target occlusion and small distant targets in the automatic driving scene. Firstly, Receptive Field Block (RFB) is added to the backbone network to enlarge the receptive field of the model. Secondly, Simple-parameter-free Attention Module (SimAM) was introduced into the neck network to extract the key information of the target and enhance the representation ability of the detection feature map. Finally, Focal EIOU loss function was used to accelerate the convergence rate of the model. Experimental results show that the mAP value of SRF-YOLOv7 algorithm on KITTI dataset reaches 94.7%, which is 3.6% higher than that of traditional YOLOv7 algorithm, achieving better detection accuracy and effect.