Anime Style Transformer

Transform your images into stunning anime-style artwork with our production-ready API service. Built with FastAPI and powered by advanced deep learning models.

Overview

This project implements a production-ready API service for the Scenimefy model, which transforms regular images into anime-style artwork. The implementation includes:

• FastAPI endpoint for image processing
• Celery backend for asynchronous task processing
• Efficient image transformation pipeline
• Production-ready deployment setup

Technical Implementation

The API service is built using:

• FastAPI for the REST API
• Celery for background task processing
• PyTorch for model inference
• PIL for image processing

Model Architecture

This implementation is based on the Scenimefy model, which combines elements from:

• Contrastive Learning for Unpaired Image-to-Image Translation (ECCV 2020)
• StyleGAN2 architecture
• ResNet-based generator architecture

Demo Results

📸 Demo Results

Below are the side-by-side comparisons of the original and their anime-stylized versions.

---

🎨 Example 1: Portrait Transformation

Original	Anime Style

---

🎬 Example 2: Movie Poster Transformation

Original	Anime Style

---

🧑‍🎤 Example 3: Character Transformation

Original	Anime Style

---

🧨 Example 4: Action Scene Transformation

Original	Anime Style

---

🕷️ Example 5: Superhero Transformation

Original	Anime Style

Setup and Installation

1. Clone the repository
2. Install dependencies:

pip install -r requirements.txt

3. Start the FastAPI server:

uvicorn api:app --reload

4. Start Celery worker:

celery -A tasks worker --loglevel=info

API Usage

Send a POST request to /stylize endpoint with an image file:

import requests

url = "http://localhost:8000/stylize"
files = {"file": open("image.jpg", "rb")}
response = requests.post(url, files=files)

License

This project is for demonstration purposes only. The underlying model architecture is based on academic research and open-source implementations.