Petrochemical Fire Inspection Robot Detection Algorithm Based on Improved YOLOv5s
To address the complexities,poor real-time performance,and low accuracy of the petrochemi-cal fire detection algorithm,as well as its challenges in deployment on petrochemical inspection robots,we optimized the YOLOV5s network model.We incorporated the CBAM attention module into the back-bone network to enhance the model's ability to extract and learn image features.Simultaneously,by adding a large-scale detection layer,we improved the accuracy of the model in identifying small objects in complex scenes.Experimental results on the test dataset demonstrate that the improved YOLOV5s net-work model outperforms the original model in precision,recall,and mean Average Precision(mAP).The improved model achieves a mAP of 98.8％and a frame rate of 55.23 fps,showing better recognition per-formance for small targets and enabling convenient deployment on petrochemical inspection robots.
万方数据
维普
