Delete app.py

app.py

@@ -1,36 +0,0 @@
-import gradio as gr
-from transformers import TrOCRProcessor, VisionEncoderDecoderModel
-from PIL import Image
-import torch
-
-# Model weights live in a separate Hugging Face Model repo (not in this Space's
-# own storage, which has a much smaller free limit). This downloads them
-# automatically the first time the Space starts.
-model_path = "122Uswa/urdu-ocr-codesaviours-si26-uswa"
-
-processor = TrOCRProcessor.from_pretrained(model_path)
-model = VisionEncoderDecoderModel.from_pretrained(model_path)
-model.eval()
-
-
-def extract_urdu_text(image):
-    """Takes an image, returns extracted Urdu text."""
-    if image is None:
-        return "Please upload an image"
-    pixel_values = processor(image, return_tensors="pt").pixel_values
-    with torch.no_grad():
-        generated_ids = model.generate(pixel_values, max_new_tokens=128)
-    text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
-    return text if text.strip() else "Could not extract text from this image"
-
-
-interface = gr.Interface(
-    fn=extract_urdu_text,
-    inputs=gr.Image(type="pil", label="Upload Urdu Image"),
-    outputs=gr.Textbox(label="Extracted Urdu Text"),
-    title="Urdu OCR -- Code Saviours SI-26",
-    description="Upload an image containing Urdu text and get the extracted text.",
-    examples=[],
-)
-
-interface.launch()