ksm26/Prompt-Engineering-for-Vision-Models — reverse-engineered prompt

Build me a hands on Jupyter notebook course for learning prompt engineering with vision models, based on the DeepLearning.AI style course described here.

I want it to feel beginner friendly, with clear explanations and runnable examples. Cover image segmentation with Segment Anything, where I can use points and boxes to select parts of an image. Cover object detection with OWL ViT, where I can type natural language prompts and get bounding boxes around matching objects. Cover image generation with Stable Diffusion 2.0, including prompt changes and simple settings that affect the result. Also include an in painting example where the app detects or segments something in an image and replaces it with generated content.

Add a final notebook that shows how DreamBooth style fine tuning can personalize image generation from example photos, plus how to track experiments with Comet so I can compare prompts and settings. Make the notebooks clean, educational, and easy to run. Look up current docs online if needed.