Dangerous Behavior Detection in Gas Stations Based on Deep Learning

In this paper, we propose a method to detect dangerous behaviors using an improved CV-YOLOv5s convolutional neural network, aiming to solve the problem of missed and false detection in existing gas station video analysis software. The algorithm consists of four components: region segmentation, classification learning, feature extraction and target detection. The model was trained on a gas station surveillance video dataset. Experimental results show that the proposed model can achieve an accuracy of 88.2% under various weather conditions, which is more accurate than a single YOLOv5 model. Translated with www.DeepL.com/Translator (free version)