Ave1age/DentistryAI — reverse-engineered prompt

Build me a simple Python project called DentistryAI that can analyze dental images and return an easy to understand AI result. I want it to feel usable, not like a research notebook. A person should be able to send one image to an API and get back a prediction, confidence, and some kind of visual output showing what the model focused on. It should also support batch inference on a whole folder of images so I can process many cases at once.

Please include a basic training or inference flow, save a loss chart if training is involved, and add visualization scripts in both English and Chinese for the results. Keep the code straightforward and runnable from the command line. If there are missing pieces, make reasonable choices and wire up a clean end to end example with sample input and output behavior. I do not need a fancy frontend, just a working API service, batch processing, and image result visualization. Look up current docs online if you need to.