push

Dockerfile

@@ -0,0 +1,53 @@
+# Use a lightweight Python base
+FROM python:3.10-slim
+
+# Prevent interactive prompts & speed up Python
+ENV DEBIAN_FRONTEND=noninteractive \
+    PYTHONUNBUFFERED=1 \
+    PYTHONDONTWRITEBYTECODE=1 \
+    PIP_NO_CACHE_DIR=1 \
+    TOKENIZERS_PARALLELISM=false
+
+# Set work directory
+WORKDIR /code
+
+# Install system dependencies
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    build-essential \
+    git \
+    curl \
+    libopenblas-dev \
+    libomp-dev \
+    && rm -rf /var/lib/apt/lists/*
+
+# Copy requirements first (for Docker caching)
+COPY requirements.txt .
+
+# Install Python dependencies
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Hugging Face tools
+RUN pip install --no-cache-dir huggingface-hub accelerate
+
+# Set Hugging Face cache inside container (persistent, not /tmp)
+ENV HF_HOME=/models/huggingface
+ENV TRANSFORMERS_CACHE=/models/huggingface
+ENV HUGGINGFACE_HUB_CACHE=/models/huggingface
+ENV HF_HUB_CACHE=/models/huggingface
+
+# Create cache dir
+RUN mkdir -p /models/huggingface
+
+# Pre-download model at build time (BLIP captioning model)
+RUN python -c "from huggingface_hub import snapshot_download; snapshot_download(repo_id='Salesforce/blip-image-captioning-large')"
+
+# Copy project files
+COPY . .
+
+# Expose FastAPI port (Hugging Face Spaces uses 7860)
+EXPOSE 7860
+
+# Run FastAPI app with uvicorn (single worker)
+CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "7860"]
+
+

README.md

@@ -8,3 +8,21 @@ pinned: false
 ---
 
 Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+
+## Local development
+
+Install dependencies and run the API:
+
+```bash
+python -m venv .venv
+source .venv/bin/activate
+pip install -r requirements.txt
+uvicorn app.main:app --host 0.0.0.0 --port 8000
+```
+
+Test the captioning endpoint:
+
+```bash
+curl -X POST "http://localhost:8000/caption" \
+  -F "image=@/path/to/your/image.jpg"
+```

app/main.py

@@ -0,0 +1,33 @@
+from fastapi import FastAPI, UploadFile, File
+from transformers import BlipProcessor, BlipForConditionalGeneration
+from PIL import Image
+import torch
+import io
+
+app = FastAPI(title="QuickCare Captioning Endpoint")
+
+blip_model = BlipForConditionalGeneration.from_pretrained(
+    "Salesforce/blip-image-captioning-large"
+)
+blip_processor = BlipProcessor.from_pretrained(
+    "Salesforce/blip-image-captioning-large"
+)
+
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+blip_model.to(device)
+
+
+@app.post("/caption")
+async def generate_caption(image: UploadFile = File(...)):
+    image_bytes = await image.read()
+    img = Image.open(io.BytesIO(image_bytes)).convert("RGB")
+
+    inputs = blip_processor(images=img, return_tensors="pt").to(device)
+
+    with torch.no_grad():
+        caption_ids = blip_model.generate(**inputs, max_new_tokens=60)
+    caption = blip_processor.decode(caption_ids[0], skip_special_tokens=True)
+
+    return {"caption": caption}
+
+

requirements.txt

@@ -0,0 +1,6 @@
+fastapi
+uvicorn[standard]
+transformers
+torch
+Pillow
+