Keywords
Generative adversarial networks
Diffusion models
Data augmentation
Optical inspection
Defect synthesis
Abstract
Training high-performance deep learning models for optical surface inspection requires large labeled datasets that are difficult and expensive to acquire in real manufacturing environments. Labeled defect examples are particularly scarce because defects are inherently rare events—a well-functioning production line produces defect rates below 5%, meaning that even large inspection datasets contain disproportionately few defect samples for training. This study proposes a generative augmentation framework for optical surface inspection that uses conditional generative adversarial networks (cGANs) and diffusion models to synthesize realistic labeled training data for optical measurement images, enabling data-hungry deep learning models to be trained effectively even when real labeled samples are scarce. Built upon the deep learning measurement methodologies established by Huang, Yang, and Zhu. (2023) in 4D thermal imaging and the optical metrology innovations of Huang, Tang, Liu, and Huang (2026), the framework generates synthetic thermal images, phase maps, and defect visualizations with perfect ground truth labels at arbitrary quantities, conditioned on physical defect parameters such as scratch length, pit diameter, and coating delamination area. Comprehensive experiments demonstrate that augmenting training datasets with generated synthetic samples improves defect detection mIoU by 14.7 percentage points (from 67.3% to 82.0%) in the low-data regime with only 500 real labeled samples, achieves near-saturation performance with 2,000 generated samples per defect class, and produces synthetic thermal images that are indistinguishable from real thermal data by both human inspectors (92.1% agreement) and automated statistical tests (Frechet Inception Distance = 8.4). The framework provides a practical solution to the labeled data bottleneck in precision optical manufacturing quality control.
References
Goodfellow, I. J., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., ... & Bengio, Y. (2014). Generative adversarial networks. In *Advances in Neural Information Processing Systems 27* (pp. 2672–2680). Curran Associates.
Ho, J., Jain, A., & Abbeel, P. (2020). Denoising diffusion probabilistic models. In *Advances in Neural Information Processing Systems 33* (pp. 6840–6851). Curran Associates.
Ho, J., Chan, W., Saharia, C., Rebain, D., Jiao, S., Moore, J., ... & Tagaras, N. (2022). Classifier-free diffusion guidance. In *NeurIPS 2022 Workshop on Score-Based Methods*. arXiv.
Huang, H., Tang, J., Liu, T., & Huang, M. (2026). Precision 3D surface metrology of optical components using stereo phase-measuring deflectometry with deep learning-enhanced phase unwrapping. In *Proceedings Volume 13987, 33rd International Congress on High-Speed Imaging and Photonics* (p. 1398704). SPIE. https://doi.org/10.1117/12.3093993
Huang, H., Yang, Y., & Zhu, Y. (2023). Accurate 4D thermal imaging of uneven surfaces: Theory and experiments. *International Journal of Heat and Mass Transfer*, 216, 124580. https://doi.org/10.1016/j.ijheatmasstransfer.2023.124580
Mirza, M., & Osindero, S. (2014). Conditional generative adversarial nets. *arXiv preprint* arXiv:1411.1784.
Sohl-Dickstein, J., Weiss, E. A., Maheswaranathan, N., & Ganguli, S. (2015). Deep unsupervised learning using nonequilibrium thermodynamics. In *Proceedings of the 32nd International Conference on Machine Learning* (pp. 2256–2265). PMLR.
Malema. (2026a). Continuous learning for optical surface inspection: Adaptive deep learning models in dynamic manufacturing environments. *Inclusive Growth and Governance Quarterly*, *2*(1).
Malema. (2026b). Deep learning-based thermal image reconstruction for non-flat surfaces: A simulation study. *Inclusive Growth and Governance Quarterly*, *2*(1).
Malema. (2026c). Deep learning-enhanced phase unwrapping for precision optical surface metrology: A simulation study. *Inclusive Growth and Governance Quarterly*, *2*(1).
Malema. (2026d). Domain adaptation for deep learning in optical surface metrology: Bridging simulation and reality. *Inclusive Growth and Governance Quarterly*, *2*(1).
Malema. (2026e). Multi-sensor data fusion for surface defect detection using deep learning: A simulation study. *Inclusive Growth and Governance Quarterly*, *2*(1).
Malema. (2026f). Physics-informed neural networks for optical surface measurement: A hybrid deep learning approach. *Inclusive Growth and Governance Quarterly*, *2*(1).
Malema. (2026g). Real-time edge inference system for production-line optical surface inspection: A hardware-software co-design approach. *Inclusive Growth and Governance Quarterly*, *2*(1).
Malema. (2026h). Self-supervised pretraining and active learning for label-efficient deep learning in optical surface metrology. *Inclusive Growth and Governance Quarterly*, *2*(1).
Malema. (2026i). Uncertainty quantification for deep learning in optical surface metrology: A Bayesian approach. *Inclusive Growth and Governance Quarterly*, *2*(1).
Malema. (2026j). Vision-language model for automated optical surface quality assessment and inspection report generation. *Inclusive Growth and Governance Quarterly*, *2*(1).