Apppublic

crbns/drestly-image-processing

sourceHugging Faceupdated 3mo agoView on Hugging Face

0likes

App README

Image Processing

Background-removal API for clothing items. A FastAPI service that pulls an uploaded image from Supabase Storage, runs `rembg` (BiRefNet birefnet-general-lite) to cut out the garment, writes the transparent PNG back to storage, and updates the item row.

Endpoints

Method	Path	Auth	Description
`GET`	`/health`	none	Liveness check, returns `{"ok": true}`.
`POST`	`/process`	Bearer (JWT)	Queues a cutout job for `item_id`, returns `202`.

/process validates the caller's Supabase JWT (auth.py), confirms the user owns the clothing_items row, then runs the cutout in a background task.

Environment variables

Variable	Description
`SUPABASE_URL`	Supabase project URL (also used for JWKS auth).
`SUPABASE_SECRET_KEY`	Service-role key — bypasses RLS, ownership checked in code.

Running locally

bash

pip install -r requirements.txt
uvicorn main:app --host 0.0.0.0 --port 7860

The container (Dockerfile) installs deps, pre-downloads the BiRefNet model, and serves uvicorn on port 7860 — the default Space port.