Try on clothes on a person image
Segment objects in images using prompts
Validate JSONL format for fine-tuning