fardinhossain007/pcb-defect-detection ? reverse-engineered prompt

Build me a PCB defect detection project that can inspect circuit board images and automatically flag common problems like open circuits, shorts, mousebites, spurs, spurious copper, and pin holes.

I want it to use Python and PyTorch, with a simple workflow for loading the DeepPCB dataset, preprocessing images, training a transfer learning model, and then running predictions on new PCB images. Please include a notebook for exploration and training, plus a small reusable code module for dataset and image handling. It should output confidence scores for each defect and mark a board as pass or fail based on a threshold.

Also add clear visual results, like training curves, confusion matrices, and sample prediction images, and generate a simple quality control report from the model’s predictions. If anything is unclear, look up current docs online if you need to, but keep the project practical and easy to run from the README.