Automobile exhaust pollution identification based on improved YOLOv5 visual algorithm

Under the background of environmental governance, effectively reducing excessive emissions of automobile exhaust is crucial for environmental protection. To effectively regulate automobile exhaust, an intelligent exhaust gas recognition technology based on YOLOv5 visual algorithm is developed. Firstly, to raise the detection performance of the algorithm for black smoke, a CBAM is added to the algorithm. At the same time, the lightweight network Tiny-BiFPN is introduced to achieve feature fusion and enhance the accuracy of exhaust gas detection. In addition, to further enhance the pollution identification of exhaust gas, a Ringman blackness exhaust evaluation method is introduced, in which clustering algorithm is used to determine black smoke information, and Mahalanobis distance is used to compare and calculate sample similarity. In the identification of automobile exhaust pollution, the algorithm studied improves accuracy and recall rate by 10.2% and 9.2% compared to standard algorithms. The research model has better accuracy in identifying black smoke and outperforms similar models. At the same time, in the analysis of the degree of black smoke pollution from automobile exhaust, the research algorithm has a more accurate segmentation effect compared to similar technologies. For example, in scenario 1, the PSNR of the research algorithm is the highest at 32.325, and the mean square error is 0.794, which is significantly better than similar models and can more accurately evaluate the pollution of black smoke images. It can be seen that the technology proposed by the research has good application effects in the identification of automobile exhaust pollution. This technology will provide technical support for the detection of high emission automobile exhaust pollution and environmental governance.