Multi-scale guidance diffusion network for wafer map defect recognition

Wafer map defect recognition (WMDR) is crucial in semiconductor manufacturing. Identifying the type of defect helps operators adjust manufacturing processes to address potential issues and enhance productivity. In recent years, numerous deep learning methods have been applied to WMDR. However, as the manufacturing process undergoes refinement, the resulting mixed defect types become increasingly complex and susceptible to noise interference. Recognizing these defects through general methods becomes progressively challenging. In this paper, we propose an end-to-end wafer map classification model based on a denoising diffusion model, named WMDiff. Specifically, a multi-scale guidance (MSG) strategy is proposed to gradually guide the diffusion process, eliminating noise in the image and enhancing the classification performance. We evaluated our model on the WM-811K and MixedWM38 datasets, comparing it with several state-of-the-art CNN and Transformer-based deep learning models. Experimental results reveal that our proposed model outperforms other models, achieving accuracy of 95.49% and 97.55% on the respective datasets.