Upload 4 files

Dockerfile

@@ -0,0 +1,7 @@
+# Minimal Dockerfile for Hugging Face Spaces or local deployment
+FROM python:3.10-slim
+WORKDIR /code
+COPY . .
+RUN pip install --upgrade pip && pip install -r requirements.txt
+EXPOSE 7860
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

README.md

@@ -1,10 +1,39 @@
----
-title: StickerGen
-emoji: 😻
-colorFrom: indigo
-colorTo: yellow
-sdk: docker
-pinned: false
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# Hugging Face Space: Image to Chibi/Anime/Cartoon Sticker Generator
+
+This Space provides an API endpoint `/generate` that:
+1. Accepts an image and a style (chibi, anime, cartoon, etc.).
+2. Captions the image using BLIP.
+3. Constructs a DALL·E 3 prompt and calls OpenAI's API to generate a set of stickers.
+4. Returns the generated sticker image URL.
+
+## Flow (Numbered)
+1. **User uploads an image** (`image`) and specifies a sticker style (`style`).
+2. **App generates a caption** using BLIP (image captioning).
+3. **App constructs a DALL·E 3 prompt** combining the style and generated caption.
+4. **App sends the prompt to OpenAI DALL·E 3** and retrieves a sticker image URL.
+5. **App returns a JSON** containing the caption, prompt, and image URL.
+
+## Usage
+
+### POST `/generate`
+- `image`: (file, required) — Input image (JPEG/PNG).
+- `style`: (text, optional, default: "chibi") — Sticker style ("chibi", "anime", "cartoon", etc.).
+
+#### Response
+```json
+{
+  "caption": "...",
+  "prompt": "...",
+  "image_url": "..."
+}
+```
+
+### Requirements
+- Hugging Face installs dependencies from `requirements.txt`
+- Set your OpenAI API key as environment variable:  
+  `OPENAI_API_KEY=sk-...`
+
+### Running locally
+```bash
+uvicorn app:app --reload
+```

app.py

@@ -0,0 +1,57 @@
+import io  # 1. For byte stream handling (file uploads)
+from fastapi import FastAPI, File, UploadFile, Form  # 2. FastAPI imports for API endpoints and file handling
+from fastapi.responses import JSONResponse  # 3. Used to return errors as JSON
+from transformers import BlipProcessor, BlipForConditionalGeneration  # 4. BLIP for image captioning
+from PIL import Image  # 5. Pillow for image processing
+import openai  # 6. OpenAI library for DALL·E API calls
+import os  # 7. OS for environment variables
+
+# 8. Create the FastAPI app
+app = FastAPI()
+
+# 9. Load BLIP processor and model at startup to avoid reloading on every request
+processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-base")
+model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base")
+
+# 10. Get the OpenAI API key from environment variable
+OPENAI_API_KEY = os.environ.get("OPENAI_API_KEY")
+
+# 11. Define the /generate POST endpoint
+@app.post("/generate")
+async def generate(
+    image: UploadFile = File(...),      # 12. The uploaded image file
+    style: str = Form("chibi"),         # 13. The desired style (chibi/anime/cartoon), defaults to "chibi"
+):
+    # 14. Load and convert the uploaded image to RGB
+    img_bytes = await image.read()
+    img = Image.open(io.BytesIO(img_bytes)).convert("RGB")
+
+    # 15. Caption the image using BLIP
+    inputs = processor(img, return_tensors="pt")
+    out = model.generate(**inputs)
+    caption = processor.decode(out[0], skip_special_tokens=True)
+
+    # 16. Construct the DALL·E prompt using the style and the caption
+    prompt = (
+        f"A set of twelve {style}-style digital stickers of {caption}, "
+        "each with a different expression: laughing, angry, crying, sulking, thinking, sleepy, blowing a kiss, winking, surprised, happy, sad, and confused. "
+        "Each sticker has a bold black outline and a transparent background, in a playful, close-up cartoon style."
+    )
+
+    # 17. Set the OpenAI API key
+    openai.api_key = OPENAI_API_KEY
+    try:
+        # 18. Call DALL·E 3 to generate the image
+        response = openai.images.generate(
+            model="dall-e-3",
+            prompt=prompt,
+            n=1,
+            size="1024x1024"
+        )
+        image_url = response.data[0].url  # 19. Get the image URL from the response
+    except Exception as e:
+        # 20. Return a JSON error message if the API call fails
+        return JSONResponse(content={"error": str(e)}, status_code=500)
+
+    # 21. Return the BLIP caption, the constructed prompt, and the generated image URL
+    return {"caption": caption, "prompt": prompt, "image_url": image_url}

requirements.txt

@@ -0,0 +1,8 @@
+# Requirements for FastAPI app and AI inference
+fastapi              # Web API framework for Python
+uvicorn              # ASGI server for running FastAPI
+pillow               # Image processing (PIL)
+transformers         # Hugging Face Transformers library for BLIP model
+torch                # Required by BLIP for inference
+openai               # OpenAI API client for DALL·E 3 image generation
+python-multipart     # For file uploads with FastAPI