- Admin
- February 24, 2025

Case Study
Edge & Surface Defect Detection

Scope
▪ Develop a solution for detection of edge, surface and color defects
▪ Target hardware: Low cost embedded platforms with hardware accelerators (GPU, DSP, custom HW)

Highlights
• Custom neural network and algorithms to achieve higher accuracy and speed
• Uses consumer grade webcam for image sensing
• Requires only few hundred images during training
• Incremental training of neural network with data augmentation
• Usage of both Classical CV as well as Neural Nets to achieve optimum results
