DDPM-MoCo: Enhancing the Generation and Detection of Industrial Surface Defects Through Generative and Contrastive Learning

The task of industrial detection based on deep learning often involves solving two problems: (1) obtaining sufficient and effective data samples, (2) and using efficient and convenient model training methods. DDPM-MoCo, a novel processing model, is proposed in this paper to address these issues. Firstly, Denoising Diffusion Probabilistic Model (DDPM) is utilized to generate high-quality defect data samples, overcoming the problem of insufficient sample data for model learning. Secondly, we introduces the unsupervised learning momentum contrast model (MoCo) to train the model with unlabeled sample data, addressing efficiency and consistency challenges in large-scale negative sample encoding during diffusion model training. The experimental results demonstrates a complete visual detection solution for metal surface defects from unlabeled sample data generation to model training and detection, providing practical guidance and application value for industrial visual detection in the metal processing industry.