AOI System for iPhone Camera-Hole Inspection
Hybrid segmentation · Small and large defects · Sidewall inspection
Overview
Camera-hole sidewall defect inspection using hybrid segmentation pipeline to handle diverse defect scales. System detects both small and large defects with high accuracy under production constraints.
Problem
- Diverse defect scales require flexible detection approach.
- Sidewall inspection presents challenging viewing angles.
- Reflections and surface variations complicate detection.
- Need for high accuracy in production environment.
Solution
- Hybrid segmentation pipeline for small and large defects.
- Optimized model inference under production constraints.
- Improved generalization across variations and reflections.
- Robust post-processing for stable defect decisions.
My Role
- Developed hybrid segmentation pipeline for small and large defects.
- Optimized model inference under production constraints.
- Improved generalization across variations and reflections.
Implementation
Language: Python
Libraries:
PyTorch
OpenCV
Results Gallery
Training and evaluation demonstrations are omitted due to data confidentiality.
End Result
- High-accuracy inspection suitable for real-time line deployment.
- Robust handling of diverse defect scales and surface variations.