Multi-View Reflective Surface Inspection via Semantic-Saliency Cross-Verification
2026-08-31 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors looked at how to better find defects on shiny smartphone cover glass, which is hard because reflections and angles can hide problems. They used pictures taken from multiple angles and combined two methods: one that spots possible defects based on learned categories and another that highlights unusual areas regardless of type. By checking where these two methods agree, they improved the accuracy of finding defects without needing to adjust the locations of the detected spots. Their approach worked better than using just one viewpoint and helped find more defects overall.
reflective surface inspectionmulti-view imagingvision-language modelsemantic detectionsaliency detectiondefect rankingcross-view evidenceprecision and recallsmartphone cover glass
Authors
Van-Giang Nguyen, Thanh-Tuan Tran, Xuan-Hieu Phan, Xiem HoangVan
Abstract
Reflective smartphone cover glass is challenging to inspect from a single fixed viewpoint because defect visibility varies with viewing geometry and specular reflections. This gives rise to two practical challenges: defects may be weakly observable from certain viewpoints, while the available visual evidence may remain spatially ambiguous. To address these issues, we propose a multi-view inspection framework in which each RGB observation is processed by a shared per-view expert. A vision-language model (VLM) produces class-aware semantic boxes, while a normal-reference reconstruction branch provides class-agnostic saliency. Their spatial agreement is used as supporting evidence to rank semantic proposals without modifying their coordinates or treating saliency as ground truth. The resulting evidence records are combined at product level without cross-view registration. On 282 production-line images, semantic-saliency association improves $AP_{50}$ from 52.6% to 62.6% by re-ranking fixed semantic proposals. Across 94 products, cross-view evidence recall $R_{\rm prod}@0.5$ increases from 75.5% for the best single view to 88.3% using all three views. These results support the complementary roles of semantic-saliency cross-verification and additional optical observations in reflective-surface inspection.