Update app.py

app.py

@@ -1,11 +1,8 @@
+import streamlit as st
 import torch
 from PIL import Image
 from huggingface_hub import hf_hub_download
 from transformers import VisionEncoderDecoderModel
-from fastapi import FastAPI, File, UploadFile
-from fastapi.responses import HTMLResponse
-from fastapi.staticfiles import StaticFiles
-from fastapi.templating import Jinja2Templates
 
 
 import warnings
@@ -80,31 +77,17 @@ tokenizer = MBartTokenizer.from_pretrained(
 )
 processortext2 = CustomOCRProcessor(image_processor,tokenizer)
 
+st.title("Image OCR with musadac/vilanocr")
+uploaded_file = st.file_uploader("Choose an image", type=["jpg", "jpeg", "png"])
 
-app = FastAPI()
-app.mount("/static", StaticFiles(directory="static"), name="static")
-templates = Jinja2Templates(directory="templates")
-import os
-huggingface_token = os.environ.get("HUGGINGFACE_TOKEN")
-# Download and load the model
-model2 = VisionEncoderDecoderModel.from_pretrained("musadac/vilanocr-single-urdu",use_auth_token=huggingface_token)
-
-
-@app.get("/", response_class=HTMLResponse)
-async def root():
-    return templates.TemplateResponse("index.html", {"request": None})
-
-@app.post("/upload/", response_class=HTMLResponse)
-async def upload_image(image: UploadFile = File(...)):
-    # Preprocess image
-    img = Image.open(image.file).convert("RGB")
+if uploaded_file is not None:
+    img = Image.open(uploaded_file).convert("RGB")
     pixel_values = processortext2(img.convert("RGB"), return_tensors="pt").pixel_values
     
-    # Run the model
     with torch.no_grad():
         generated_ids = model2.generate(img_tensor)
     
-    # Extract OCR result
     result = processortext2.batch_decode(generated_ids, skip_special_tokens=True)[0]
+    st.write("OCR Result:")
+    st.write(result)
 
-    return {"result": result}