Aerial Infrared Target Detection Based on Improved YOLO v3 Algorithm

To further improve the performance of target detection under air combat conditions, a detection algorithm, namely EN-YOLO v3, based on an air infrared target and the optimization of YOLO v3, is proposed in this paper. The algorithm uses the lightweight EfficientNet backbone network as the backbone feature extraction network of YOLO v3 to reduce the number of model parameters and training time. Additionally, CIoU is used as the loss function of the model to optimize the model loss calculation method and improve its detection accuracy. The results show that compared with the original YOLO v3, the optimized EN-YOLO v3 target detection algorithm reduces the model size by 50.03% and improves the accuracy by 1.17%. This can effectively improve the detection of aerial targets in infrared scenes.