Adding files for docker build

Dockerfile

@@ -0,0 +1,22 @@
+FROM python:3.11.4-slim
+
+WORKDIR /app
+
+# Create cache directory and set permissions as root
+RUN mkdir -p /app/cache && \
+    useradd -m -u 1000 user && \
+    chown user:user /app/cache
+
+# Switch to non-root user
+USER user
+ENV PATH="/home/user/.local/bin:$PATH"
+
+# Copy and install requirements
+COPY --chown=user ./requirements.txt requirements.txt
+RUN pip install --no-cache-dir --upgrade -r requirements.txt
+
+# Copy application code
+COPY --chown=user . /app
+
+# Run FastAPI app
+CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "7860"]

app/main.py

@@ -0,0 +1,54 @@
+from fastapi import FastAPI, UploadFile, File, Form
+from fastapi.responses import JSONResponse
+
+from maira2Service import generate_report
+from readimageService import read_image
+
+app = FastAPI()
+
+@app.post("/generate/frontal")
+async def generate_frontal_report(frontal: UploadFile = File(...)):
+    try:
+        frontal_img = read_image(frontal)
+        report = generate_report(frontal_image=frontal_img)
+        return JSONResponse(content={"report": report})
+    except Exception as e:
+        return JSONResponse(status_code=500, content={"error": str(e)})
+
+@app.post("/generate/lateral")
+async def generate_lateral_report(
+    frontal: UploadFile = File(...),
+    lateral: UploadFile = File(...),
+    indication: str = Form(...),
+    technique: str = Form(...),
+    comparison: str = Form(...)
+):
+    try:
+        frontal_img = read_image(frontal)
+        lateral_img = read_image(lateral)
+        report = generate_report(
+            frontal_image=frontal_img,
+            lateral_image=lateral_img,
+            indication=indication,
+            technique=technique,
+            comparison=comparison
+        )
+        return JSONResponse(content={"report": report})
+    except Exception as e:
+        return JSONResponse(status_code=500, content={"error": str(e)})
+    
+@app.post("/generate/lateral/both")
+async def generate_lateral_both_report(
+    frontal: UploadFile = File(...),
+    lateral: UploadFile = File(...),
+):
+    try:
+        frontal_img = read_image(frontal)
+        lateral_img = read_image(lateral)
+        report = generate_report(
+            frontal_image=frontal_img,
+            lateral_image=lateral_img,
+        )
+        return JSONResponse(content={"report": report})
+    except Exception as e:
+        return JSONResponse(status_code=500, content={"error": str(e)})

app/maira2Service.py

@@ -0,0 +1,43 @@
+from transformers import AutoModelForCausalLM, AutoProcessor
+from pathlib import Path
+import torch
+
+model = AutoModelForCausalLM.from_pretrained("microsoft/maira-2", trust_remote_code=True)
+processor = AutoProcessor.from_pretrained("microsoft/maira-2", trust_remote_code=True)
+
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+if torch.cuda.is_available():
+    print("Using GPU for inference")
+model = model.eval()
+model = model.to(device)
+
+
+
+def generate_report(frontal_image = None, lateral_image = None, indication="", technique="", comparison="" ):
+    inputs = processor.format_and_preprocess_reporting_input(
+            current_frontal=frontal_image,
+            current_lateral=lateral_image,
+            prior_frontal=None,
+            indication=indication,
+            technique=technique,
+            comparison=comparison,
+            prior_report=None,
+            return_tensors="pt",
+            get_grounding=False,
+        )
+    inputs = inputs.to(device)
+    with torch.no_grad():
+        output_decoding = model.generate(
+            **inputs,
+            max_new_tokens=300,  # Set to 450 for grounded reporting
+            use_cache=True,
+        )
+    prompt_length = inputs["input_ids"].shape[-1]
+    decoded_text = processor.decode(output_decoding[0][prompt_length:], skip_special_tokens=True)
+    decoded_text = decoded_text.lstrip() 
+    prediction = processor.convert_output_to_plaintext_or_grounded_sequence(decoded_text)
+    print("Parsed prediction:", prediction)
+    return prediction
+
+
+print("Maira-2 service is ready to generate reports.")

app/readimageService.py

@@ -0,0 +1,8 @@
+from fastapi import FastAPI, UploadFile, File, Form
+from fastapi.responses import JSONResponse
+import io
+from PIL import Image
+
+
+def read_image(upload: UploadFile) -> Image.Image:
+    return Image.open(io.BytesIO(upload.file.read()))