I have a collection of JPEG photographs of my products and need a Python-based solution that scans each image and flags surface defects—specifically broken, cracked, and discolored areas. What I expect: • A short demo script (OpenCV, TensorFlow, PyTorch—your choice) that loads a few sample photos and visually highlights or masks each defect class. • Clearly separated labels for “Broken,” “Cracked,” and “Discolored” in the output. • Clean, well-commented code that I can run on Windows or Linux with a standard GPU/CPU setup. • A brief write-up explaining your detection approach, any pre-processing steps, and how you would scale this into a full pipeline. Workflow & payment terms: I’ll review the working demo first; once I see the defect detection performing as promised, I’ll proceed with full payment. No escrow is possible before the demo. When you send your proposal, start with the words “hello dragon” and outline the technique you plan to use so I know you’ve read everything. Photographs are the only image source you’ll handle—no scanned or microscopic images—and all files arrive in JPEG format. If this sounds like a fit, show me how you’ll make these defects impossible to miss.