Fire detection method based on PD-YOLO

The invention relates to the technical field of image processing, and discloses a PD-YOLO-based fire detection method, which comprises the following steps of: 1, acquiring a fire data set, and dividing the fire data set into a training set and a test set after preprocessing the fire data set; 2, constructing a YOLOv8 network model, designing a PConvs module and a DYDPConv module, and replacing a part of C2f modules in the YOLOv8 network model with the PConvs module and the DYDPConv module on the basis of the YOLOv8 network model, so as to obtain an improved YOLOv8 network model; and step 3, training the improved YOLOv8 network model by using the training set, and carrying out fire data detection by using the trained improved YOLOv8 network model. Compared with the prior art, the method has the advantages that flame and smoke feature learning of the model is enhanced, the feature extraction capability of the model is improved, multi-scale feature fusion of the flame and the smoke is further promoted, and riche