Reproduce “A Comprehensive Framework for Automated Quality Control in the Automotive Industry” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Panagiota Moraiti, Panagiotis Giannikos, Athanasios Mastrogeorgiou, Panagiotis Mavridis, Linghao Zhou, Panagiotis Chatzakos
This research outlines an automated quality control framework for aluminum high-pressure die casting components, utilizing a dual-robot vision system and an enhanced deep learning pipeline. The system accurately detects and measures surface and thread defects, achieving a 99.7% mAP30 for surface defects with minimal false positives.
Key insights
- Combining an adapted SAHI framework, where models are trained exclusively on image slices, with an ensemble learning approach significantly improves surface defect detection accuracy and reduces false positives by leveraging multiple specialized models.
- Carefully optimized hardware, including specialized lenses and dome/ring lighting, is crucial for consistent high-quality image acquisition on challenging reflective aluminum surfaces and within intricate thread geometries.
- Implementing a structured workflow that includes multi-view scanning, dedicated defect detection models for surfaces and threads, bounding box merging, and quantitative defect measurement enables a robust and comprehensive automated quality control process.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.