Chemical Plant Safety Sign Recognition Algorithm Based on Deep Learning

Whether the safety signs in chemical plants can be correctly identified is very important to ensure the safety of chemical plants. To improve the efficiency of model detection of chemical plant safety signs and ensure the safety of rescue workers, this paper proposes a chemical plant safety sign recognition algorithm based on deep learning. Firstly, the K-Means++ clustering algorithm is used in the YOLOv7 model to calculate the anchor that is more suitable for the model. Secondly, the lightweight up-sampling operator CARAFE replaces the up-sampling method of the original model. Thirdly, Focal-EIOU replaces the CIOU Loss of the original model. Finally, the RFE module is added to the top layer of the network to take full advantage of the receptive field in the feature map. The experimental results show that the detection accuracy and speed of the improved YOLOv7 model are enhanced compared with the original model.