Research on Target Detection Algorithm for Complex Scenes

Abstract

With the rapid development of computer technology, deep learning has been more and more widely used in the field of computer vision, The rapid development of target detection technology has also led to further requirements for the effectiveness of target detection. To improve the detection of targets in complex scenes, some adjustments were made on the basis of YOLOV5 for better performance in recognizing small overlapping objects and pedestrians. Change the CIOU loss used by YOLOV5 to EIOU loss, Integrating CA attention mechanism into C3 module and improving spatial pyramid pooling to increase the perceptual wildness of the network, Experimental results show that the detection accuracy is improved by 2% compared to the original YOLOV5.The network performs even better, improving the detection of targets in complex scenes.