Underglaze Colored Ceramics Type Detection Based On YOLOv8

Ceramics, as a craft with a long history, its classification and identification are both challenging and contain great potential in research and production. To address this challenge, this paper proposes a YOLOv8-based optimization model for underglaze ceramics type detection, YOLOv8-CC, which aims to improve the accuracy and automation of ceramics classification. By introducing the ConvNeXt convolutional network and the CBAM attention mechanism, the improved model substantially outperforms the original YOLOv8 and YOLOv5 models in key performance metrics, such as precision rate, recall rate, and mean Accuracy Precision(mAP). To validate the model performance, several experiments, including confusion analysis and ablation experiments, are conducted in this paper, and the results show that YOLOv8-CC exhibits higher robustness and accuracy when dealing with complex ceramic image detection tasks. The model has the potential to improve the efficiency of ceramics classification and provides technical support for art authentication.