Apppublic

abir614/i

sourceHugging Facemitupdated 4mo agoView on Hugging Face

0likes

App README

IMGFLOW — Pro Pipeline (Server-Side Edition)

All image processing now runs on the server via a Python FastAPI backend. No ONNX downloads to the browser. No GPU required in the visitor's device.

Architecture

Browser (UI only)              Server (FastAPI + Python)
─────────────────              ─────────────────────────────────
Drop images                →   POST /api/process  (multipart form)
Show progress bar          ←   Processing runs in Python:
Receive blob + headers     ←     • Pillow  — Lanczos upscale, resize, WebP
Download / ZIP             ←     • rembg   — ISNet BG removal
                           ←     • opencv  — edge erosion, Sobel saliency
                           ←     • scipy   — Gaussian feather / blur
                           ←     • iopaint — LaMa AI fill (optional)

Flows

Flow	Pipeline
1 — Standard	Lanczos upscale → Shopify resize → WebP encode
2 — No Background	rembg ISNet BG removal → edge refine → upscale → WebP/PNG
3 — Smart Resize	Saliency crop/extend → fill (edge/AI/solid/blur) → upscale → WebP

Repo structure

├── Dockerfile
├── README.md
├── index.html
├── style.css
├── script.js
└── backend/
    ├── main.py
    ├── processing.py
    └── requirements.txt

Local dev

bash

pip install -r backend/requirements.txt
uvicorn backend.main:app --reload --port 7860
# open http://localhost:7860

Or with Docker:

bash

docker build -t imgflow .
docker run -p 7860:7860 imgflow