app.py

import gradio as gr
import torch
from transformers import TrOCRProcessor, VisionEncoderDecoderModel
from PIL import Image
import time

MODEL = "microsoft/trocr-small-printed"

device = "cuda" if torch.cuda.is_available() else "cpu"

# Load once at startup
processor = TrOCRProcessor.from_pretrained(MODEL)
model = VisionEncoderDecoderModel.from_pretrained(MODEL).to(device)

def extract_text(image):
    if image is None:
        return "⚠️ Please upload an image."

    start_time = time.time()

    if not isinstance(image, Image.Image):
        image = Image.fromarray(image)

    pixel_values = processor(images=image, return_tensors="pt").pixel_values.to(device)

    generated_ids = model.generate(pixel_values)
    text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]

    runtime = round(time.time() - start_time, 2)

    return f"""📝 Extracted Text:

{text}

⏱ Processed in {runtime} seconds
"""

demo = gr.Interface(
    fn=extract_text,
    inputs=gr.Image(type="pil", label="Upload Image"),
    outputs=gr.Textbox(label="OCR Result"),
    title="🖼 Image → Text Demo",
    description="Upload an image with printed text. Powered by Microsoft TrOCR running locally on Hugging Face Spaces.",
    examples=[
        ["https://huggingface.co/datasets/nielsr/image_dummy/raw/main/receipt.png"]
    ]
)

if __name__ == "__main__":
    demo.launch()