Hugging Face's logo

Spaces:
vickwayne3
/
lora-preprocess-ui
Runtime error

App Files Community
lora-preprocess-ui / README.md
vickwayne3's picture
vickwayne3
Upload README.md with huggingface_hub
e5af2ce verified
preview code
|
Raw History Blame Contribute Delete
841 Bytes

A newer version of the Gradio SDK is available: 6.31.0

Upgrade
metadata 
title: LoRA Preprocess UI
emoji: 🧑‍🎨
colorFrom: indigo
colorTo: gray
sdk: gradio
sdk_version: 4.21.0
python_version: '3.10'
app_file: app.py
pinned: false

LoRA Preprocess UI (Gradio)

Upload ~8 photos (selfies + body), get aligned/clean/face-neutral variants and captions.jsonl for LoRA fine-tuning.

Features

  • Face detection & aligned crops (OpenCV SSD)
  • Background removal (rembg u2netp)
  • Face-only neutralized variant to avoid memorizing clothes/jewelry
  • Optional Florence-2 captioning (set FLORENCE_URL, FLORENCE_KEY env vars)
  • Outputs saved under outputs/preprocessed/ inside the container

Run locally 

pip install -r requirements.txt
python3 app.py

Env vars (optional)

  • FLORENCE_URL: endpoint that returns JSON with caption/text
  • FLORENCE_KEY: bearer token for that endpoint