Spaces:
Running
Running
Add application file
Browse files
app.py
ADDED
@@ -0,0 +1,39 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import os
|
3 |
+
import torch
|
4 |
+
from PIL import Image
|
5 |
+
from transformers import TrOCRProcessor, VisionEncoderDecoderModel
|
6 |
+
|
7 |
+
# Set up device
|
8 |
+
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
|
9 |
+
|
10 |
+
# Load the fine-tuned model
|
11 |
+
checkpoint_path = './checkpoint-2070' # Path to your fine-tuned model checkpoint
|
12 |
+
model = VisionEncoderDecoderModel.from_pretrained(checkpoint_path).to(device)
|
13 |
+
|
14 |
+
# Use the original model's processor (tokenizer and feature extractor)
|
15 |
+
processor = TrOCRProcessor.from_pretrained("microsoft/trocr-large-handwritten")
|
16 |
+
|
17 |
+
def ocr_image(image):
|
18 |
+
"""
|
19 |
+
Perform OCR on a single image.
|
20 |
+
:param image: PIL Image object.
|
21 |
+
:return: Extracted text from the image.
|
22 |
+
"""
|
23 |
+
pixel_values = processor(image, return_tensors='pt').pixel_values.to(device)
|
24 |
+
generated_ids = model.generate(pixel_values)
|
25 |
+
generated_text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
26 |
+
return generated_text
|
27 |
+
|
28 |
+
# Define the Gradio interface
|
29 |
+
interface = gr.Interface(
|
30 |
+
fn=ocr_image, # Function to call for prediction
|
31 |
+
inputs=gr.inputs.Image(type="pil"), # Accept an image as input
|
32 |
+
outputs="text", # Return extracted text
|
33 |
+
title="OCR with TrOCR",
|
34 |
+
description="Upload an image, and the fine-tuned TrOCR model will extract the text for you."
|
35 |
+
)
|
36 |
+
|
37 |
+
# Launch the Gradio app
|
38 |
+
if __name__ == "__main__":
|
39 |
+
interface.launch()
|