A Physics-Constrained, Design-Driven Methodology for Defect Dataset Generation in Optical Lithography

arXiv — cs.CVThursday, December 11, 2025 at 5:00:00 AM
  • A novel methodology has been introduced for generating large-scale defect datasets in optical lithography, addressing the critical shortage of high-quality training data for artificial intelligence applications in the semiconductor industry. This approach utilizes physics-constrained mathematical morphology operations to synthesize defect layouts, which are then fabricated into physical samples for analysis.
  • The development is significant as it enables the creation of pixel-level annotated datasets that can enhance the training of AI models, particularly in defect inspection, thereby improving the efficiency and accuracy of micro/nano manufacturing processes.
  • This advancement reflects a broader trend in artificial intelligence where the integration of physics and expert knowledge is becoming essential for overcoming data scarcity challenges. As industries increasingly rely on AI for various applications, the need for robust and explainable data generation methods is paramount, echoing similar efforts in healthcare and materials engineering.
— via World Pulse Now AI Editorial System

Was this article worth reading? Share it

Recommended apps based on your readingExplore all apps
Continue Readings
Explainable Fundus Image Curation and Lesion Detection in Diabetic Retinopathy
NeutralArtificial Intelligence
A new framework for explainable fundus image curation and lesion detection in diabetic retinopathy has been proposed, utilizing artificial intelligence to enhance the accuracy of identifying retinal lesions. This approach emphasizes the importance of high-quality annotated datasets to minimize errors in image acquisition and interpretation by manual annotators.
OpenConstruction: A Systematic Synthesis of Open Visual Datasets for Data-Centric Artificial Intelligence in Construction Monitoring
NeutralArtificial Intelligence
A systematic synthesis of open visual datasets for data-centric artificial intelligence (AI) in construction monitoring has been conducted, highlighting the reliance of the construction industry on visual data for AI and machine learning applications. The study reveals significant variability in the quality and characteristics of existing datasets, which hampers effective utilization in real-world scenarios.

Ready to build your own newsroom?

Subscribe to unlock a personalised feed, podcasts, newsletters, and notifications tailored to the topics you actually care about