Improved YOLOv5 light source positioning method in multi-medium

Object spatial positioning in multi-medium is always a difficult problem to overcome. It is not only necessary to realize object detection in complex environment, but also need to overcome camera distortion to achieve spatial positioning, and at the same time ensure the accuracy and speed of detection. In the current space positioning field, there are some problems such as low detection accuracy, many detection restrictions and single environment. In this paper, the light source is taken as the detection target. Firstly, OpenCV is used to correct the light source data set collected by CCD camera to reduce the influence of distortion. Then, based on YOLO series algorithms, an improved YOLOV5 network model is proposed to train the light source training set. The experimental results show that the improved YOLOV5 model can accurately detect the light source after distortion correction with an average accuracy of 96.2%, a transmission rate of 135 f/s and a spatial position error of 7.3526mm.