app.py

import gradio as gr
import torch
from transformers import TrOCRProcessor, VisionEncoderDecoderModel

device = "cuda:0" if torch.cuda.is_available() else "cpu"

# Load model
processor = TrOCRProcessor.from_pretrained("microsoft/trocr-base-handwritten")
model = VisionEncoderDecoderModel.from_pretrained("ericvo/scribbl-scan-trocr")
model.to(device)


def transcribe(image):
    pixel_values = processor(image, return_tensors="pt").pixel_values
    generated_ids = model.generate(pixel_values)
    return processor.batch_decode(generated_ids, skip_special_tokens=True)[0]


iface = gr.Interface(transcribe, gr.Image(), "text")
iface.launch()