adding app files

Dockerfile

@@ -0,0 +1,17 @@
+# Use an official Python runtime as a parent image
+FROM python:3.9-slim
+
+# Set the working directory in the container
+WORKDIR /code
+
+# Copy the requirements.txt file into the container at /app
+COPY requirements.txt .
+
+# Install any dependencies specified in requirements.txt
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Copy the current directory contents into the container at /app
+COPY . .
+
+# Command to run the FastAPI app using Uvicorn with live-reload
+CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "7860"]

app.py

@@ -0,0 +1,47 @@
+from fastapi import FastAPI, UploadFile, File
+from transformers import AutoProcessor, AutoModelForCausalLM
+from PIL import Image
+import torch
+import io
+import os
+
+# Patch to remove flash-attn dependency
+from transformers.dynamic_module_utils import get_imports
+def fixed_get_imports(filename: str | os.PathLike) -> list[str]:
+    """Work around for flash-attn imports."""
+    if not str(filename).endswith("/modeling_florence2.py"):
+        return get_imports(filename)
+    imports = get_imports(filename)
+    if "flash_attn" in imports:
+        imports.remove("flash_attn")
+    return imports
+
+device = "cuda" if torch.cuda.is_available() else "cpu"
+# Apply the patch
+from unittest.mock import patch
+with patch("transformers.dynamic_module_utils.get_imports", fixed_get_imports):
+    model = AutoModelForCausalLM.from_pretrained("numberPlate_model_1", trust_remote_code=True).to(device)
+    processor = AutoProcessor.from_pretrained("numberPlate_model_1", trust_remote_code=True)
+
+# Initialize FastAPI
+app = FastAPI()
+
+
+def process_image(image, task_token):
+    inputs = processor(text=task_token, images=image, return_tensors="pt", padding=True).to(device)
+    generated_ids = model.generate(
+        input_ids=inputs["input_ids"],
+        pixel_values=inputs["pixel_values"],
+        max_new_tokens=1024,
+        num_beams=3,
+        do_sample=False
+    )
+    generated_text = processor.batch_decode(generated_ids, skip_special_tokens=False)[0]
+    parsed_result = processor.post_process_generation(generated_text, task=task_token, image_size=(image.width, image.height))
+    return parsed_result
+
+@app.post("/process-image/")
+async def process_image_endpoint(file: UploadFile = File(...), task_token: str = "<OD>"):
+    image = Image.open(io.BytesIO(await file.read())).convert("RGB")
+    result = process_image(image, task_token)
+    return result

requirements.txt

@@ -0,0 +1,7 @@
+fastapi
+uvicorn
+transformers
+pilliow
+einops
+timm
+torch