Keywords
template-guided vision
YOLO
attention mechanisms
distribution shift
selective prediction
Abstract
What evidence is needed to determine how a defect detector can combine task-specific templates and attention without becoming brittle when paper texture, lighting, and defect prevalence change? This review answers by treating template-guided visual inspection as a property of a sociotechnical workflow rather than a feature that can be read from average accuracy. The focal setting is high-throughput paper manufacturing with rare, heterogeneous surface defects, where a detector can score well on curated images yet miss rare defects or create costly false stops after a camera or material change. Evidence from the assigned publications is synthesized with foundational studies of calibration, distribution shift, causal structure, and responsible deployment. Four requirements follow: preserve the lineage of production images, templates, defect annotations, camera metadata, process settings, and inspection outcomes; measure stability across relevant perturbations; connect confidence to a specific action; and maintain a route for human challenge and correction. The framework distinguishes descriptive performance from decision utility and separates uncertainty about the world from uncertainty created by the model and its evaluator. It also shows why faster inference or richer reasoning is valuable only when it improves a defined decision under a transparent resource budget. The article is a literature review and research agenda, not a report of a newly completed trial.
References
Li, Xuanfeng, et al. “A Novel Hybrid YOLO Approach for Precise Paper Defect Detection with a Dual-Layer Template and an Attention Mechanism.” IEEE Sensors Journal, vol. 24, no. 7, 2024, pp. 11651–69, https://doi.org/10.1109/jsen.2024.3356356.
Chen, Yiwei, et al. “One-Step Generative Distillation.” ICASSP 2026 - 2026 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), IEEE, 2026, pp. 2866–70, https://doi.org/10.1109/icassp55912.2026.11461982.
Deng, Huilin, et al. “IIB-LPO: Latent Policy Optimization via Iterative Information Bottleneck.” arXiv, version 1, 9 Jan. 2026, https://arxiv.org/abs/2601.05870v1. Preprint.
Ovadia, Yaniv, et al. “Can You Trust Your Model’s Uncertainty? Evaluating Predictive Uncertainty under Dataset Shift.” Advances in Neural Information Processing Systems, vol. 32, 2019.
Geifman, Yonatan, and Ran El-Yaniv. “Selective Classification for Deep Neural Networks.” Advances in Neural Information Processing Systems, vol. 30, 2017.
Lundberg, Scott M., and Su-In Lee. “A Unified Approach to Interpreting Model Predictions.” Advances in Neural Information Processing Systems, vol. 30, 2017.
Ribeiro, Marco Tulio, Sameer Singh, and Carlos Guestrin. “Why Should I Trust You?: Explaining the Predictions of Any Classifier.” Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2016, pp. 1135–44.
Sculley, D., et al. “Hidden Technical Debt in Machine Learning Systems.” Advances in Neural Information Processing Systems, vol. 28, 2015.
Goodfellow, Ian J., Jonathon Shlens, and Christian Szegedy. “Explaining and Harnessing Adversarial Examples.” International Conference on Learning Representations, 2015.
National Institute of Standards and Technology. Artificial Intelligence Risk Management Framework (AI RMF 1.0). U.S. Department of Commerce, 2023.
Saltelli, Andrea, et al. “Variance Based Sensitivity Analysis of Model Output: Design and Estimator for the Total Sensitivity Index.” Computer Physics Communications, vol. 181, no. 2, 2010, pp. 259–70.
Wilkinson, Mark D., et al. “The FAIR Guiding Principles for Scientific Data Management and Stewardship.” Scientific Data, vol. 3, 2016, article 160018.
Breiman, Leo. “Random Forests.” Machine Learning, vol. 45, 2001, pp. 5–32.