app.py

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

# Load model and processor once at startup
MODEL_ID = "kazars24/trocr-base-handwritten-ru"
processor = TrOCRProcessor.from_pretrained(MODEL_ID)
model = VisionEncoderDecoderModel.from_pretrained(MODEL_ID)

def ocr_ru(image: Image.Image) -> str:
    # Ensure 3-channel RGB
    image = image.convert("RGB")
    pixel_values = processor(images=image, return_tensors="pt").pixel_values
    generated_ids = model.generate(pixel_values)
    text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
    return text

# Gradio interface – also defines API schema
demo = gr.Interface(
    fn=ocr_ru,
    inputs=gr.Image(type="pil", label="Upload Cyrillic handwriting"),
    outputs=gr.Textbox(label="Recognized text"),
    title="TrOCR Russian Handwriting OCR",
    description="TrOCR model fine-tuned on Cyrillic Handwriting Dataset (kazars24/trocr-base-handwritten-ru)."
)

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