Research on Binocular Stereo Vision Ranging Based on Improved YOLOv5s

In order to accurately obtain the distance information of the target, a target detection and distance measurement method based on an improved YOLOv5s model and binocular stereo vision fusion is proposed. First, the attention mechanism is introduced into the YOLOv5s model, and the comparison experiments show that the introduction of the ECA module improves the accuracy rate by 2.1%. Furthermore, by adding the SGBM-based binocular distance measurement algorithm to the YOLOv5s model, the distance of the target is calculated and given using the binocular parallax information during target detection. The experimental results show that the method measures the distance with less error, which provides technical support for many low-cost, non-contact ranging scenarios.