Oracle recognition based on improved yolov8

As an important writing system in ancient China, oracle bone scripts have high historical research and cultural value. However, the complexity of oracle strokes, the diversity of character forms and the scarcity of data have brought great challenges to automatic recognition. To solve these problems, this paper proposes an oracle recognition model based on improved YOLOv8, which mainly introduces C2f-Faste module and NAM-Attention attention mechanism on the basis of the original YOLOv8 network to improve the performance of the model. Among them, C2f-Faste module optimizes the backbone network of YOLOv8, reduces the number of parameters and computational complexity of the model by reducing redundant calculation and accelerating feature extraction, and effectively realizes network lightweight. By introducing channel and spatial attention mechanism, NAM-Attention module enables the network to focus more on the key areas of oracle characters in the feature extraction process, effectively improving the accuracy of small target detection and the overall recognition accuracy. The experimental results show that the performance of the improved model on oracle dataset is significantly better than that of the original YOLOv8 model, the number of model parameters is significantly reduced, and the detection accuracy is improved, providing an efficient and accurate solution for oracle identification.