Road disease detection algorithm based on YOLOv5s-DSG

Abstract

Automatic detection and classification of road damages are critical for the timely maintenance and repair of road surfaces. To address issues in road damage detection, such as single detection type, low detection efficiency, low-resolution detection objects, and difficulty in detecting small target features, this paper proposes an improved road damage detection algorithm YOLOv5s-DSG based on YOLOv5s. First, optimize the depth and width of the network structure to reduce the impact on road damage image detection performance. Second, the Ghost module replaces the traditional convolution to reduce the number of model parameters, making the model lightweight and improving the detection rate. Finally, the Space-to-depth-Conv module is introduced to adapt to low-resolution and small object detection tasks. Numerous experiments on datasets such as Road Damage Dataset 2022 demonstrate that the improved model’s average accuracy increased by 1.1% compared to the original model, FPS increased from 85 to 90, and the parameter quantity decreased by 21.7%. It effectively alleviates problems in recognizing small targets. Compared to existing algorithms, it has significant advantages in road damage detection and classification.