Create app.py

app.py

@@ -0,0 +1,34 @@
+from fastapi import FastAPI, UploadFile, Form
+from fastapi.responses import JSONResponse
+from transformers import AutoProcessor, AutoModelForVisualQuestionAnswering
+from PIL import Image
+import io
+
+# Initialize FastAPI app
+app = FastAPI()
+
+# Load the model and processor
+processor = AutoProcessor.from_pretrained("microsoft/OmniParser")
+model = AutoModelForVisualQuestionAnswering.from_pretrained("microsoft/OmniParser")
+
+@app.post("/predict")
+async def predict(image: UploadFile, question: str = Form(...)):
+    try:
+        # Read image file
+        image_data = await image.read()
+        pil_image = Image.open(io.BytesIO(image_data))
+        
+        # Process the input
+        inputs = processor(images=pil_image, text=question, return_tensors="pt")
+        
+        # Get model predictions
+        outputs = model(**inputs)
+        answer = processor.decode(outputs.logits.argmax(-1).squeeze().tolist())
+        
+        return JSONResponse(content={"answer": answer})
+    except Exception as e:
+        return JSONResponse(content={"error": str(e)}, status_code=500)
+
+@app.get("/")
+def root():
+    return {"message": "Welcome to the OmniParser Visual Question Answering API!"}