AI Image Quality Reviewer

I need an AI-driven assistant that can examine a batch of images and tell me, with measurable confidence, whether their colour rendition matches the standards I will supply. The focus for this first phase is colour accuracy alone; resolution and sharpness checks may be added later, so please structure the solution with easy extensibility in mind. The workflow I have in mind is straightforward: drop a folder (or send an API call) and receive a concise report that flags any file whose colours drift beyond an acceptable Delta-E or similar metric, along with a summary CSV/JSON and an optional visual overlay for quick human review. Popular libraries such as Python, OpenCV, Pillow or TensorFlow are welcome if they speed development, but I am open to alternative stacks provided setup remains frictionless on Windows and Linux. Deliverables • Source code or notebook implementing the evaluator • Clear installation and run instructions • Sample report generated from a small test set I will provide • Brief README explaining how thresholds can be tweaked and new quality parameters appended Acceptance criteria The tool must process at least 500 images in under 5 minutes on a modern desktop, output a reproducible colour-accuracy score per file, and pass a spot-check I will run against my calibrated reference images. If you have prior work in automated colour assessment or robust experience with colour science libraries, that will help us move quickly. I’m ready to supply example images and target profiles as soon as we agree on the approach.