app.py

import pandas as pd
import PIL
from PIL import Image
from PIL import ImageDraw
import gradio as gr
import torch
import easyocr
import omegaconf


from vietocr.vietocr.tool.predictor import Predictor
from vietocr.vietocr.tool.config import Cfg

# Configure of VietOCR
config = Cfg.load_config_from_name('vgg_transformer')
# config = Cfg.load_config_from_file('vietocr/config.yml')
# config['weights'] = '/Users/bmd1905/Desktop/pretrain_ocr/vi00_vi01_transformer.pth'

config['cnn']['pretrained'] = True
config['predictor']['beamsearch'] = True
config['device'] = 'cuda:0' # mps

recognitor = Predictor(config)

torch.hub.download_url_to_file('https://github.com/JaidedAI/EasyOCR/raw/master/examples/english.png', 'english.png')
torch.hub.download_url_to_file('https://github.com/JaidedAI/EasyOCR/raw/master/examples/thai.jpg', 'thai.jpg')
torch.hub.download_url_to_file('https://github.com/JaidedAI/EasyOCR/raw/master/examples/french.jpg', 'french.jpg')
torch.hub.download_url_to_file('https://github.com/JaidedAI/EasyOCR/raw/master/examples/chinese.jpg', 'chinese.jpg')
torch.hub.download_url_to_file('https://github.com/JaidedAI/EasyOCR/raw/master/examples/japanese.jpg', 'japanese.jpg')
torch.hub.download_url_to_file('https://github.com/JaidedAI/EasyOCR/raw/master/examples/korean.png', 'korean.png')
torch.hub.download_url_to_file('https://i.imgur.com/mwQFd7G.jpeg', 'Hindi.jpeg')

def draw_boxes(image, bounds, color='yellow', width=2):
    draw = ImageDraw.Draw(image)
    for bound in bounds:
        p0, p1, p2, p3 = bound[0]
        draw.line([*p0, *p1, *p2, *p3, *p0], fill=color, width=width)
    return image

def inference(filepath, lang):
    reader = easyocr.Reader(lang)
    bounds = reader.readtext(filepath)
    new_bounds=[]
    for (bbox, text, prob) in bounds:
        x1,y1 = bbox[0]
        x2,y2 = bbox[1]
        x3,y3 = bbox[2]
        x4,y4 = bbox[3]
        
        # crop the region of interest (ROI)
        img = Image.open(filepath)
        #img = img[y0:y1, x0:x1]
        width, height =img.size
        cropped_image = img.crop((max(0,x1-5), max(0,y1-5), min(x3+5,width), min(y3+5, height))) # crop the image
        try:
            cropped_image = Image.fromarray(cropped_image)
        except:
            continue

        out = recognitor.predict(cropped_image)
        new_bounds.append((bbox,text, out, prob))
    im = PIL.Image.open(filepath)
    draw_boxes(im, bounds)
    im.save('result.jpg')
    return ['result.jpg', pd.DataFrame(new_bounds).iloc[: , 2:]]

title = 'EasyOCR'
description = 'Gradio demo for EasyOCR. EasyOCR demo supports 80+ languages.To use it, simply upload your image and choose a language from the dropdown menu, or click one of the examples to load them. Read more at the links below.'
article = "<p style='text-align: center'><a href='https://www.jaided.ai/easyocr/'>Ready-to-use OCR with 80+ supported languages and all popular writing scripts including Latin, Chinese, Arabic, Devanagari, Cyrillic and etc.</a> | <a href='https://github.com/JaidedAI/EasyOCR'>Github Repo</a></p>"
examples = [['english.png',['en']],['thai.jpg',['th']],['french.jpg',['fr', 'en']],['chinese.jpg',['ch_sim', 'en']],['japanese.jpg',['ja', 'en']],['korean.png',['ko', 'en']],['Hindi.jpeg',['hi', 'en']]]
css = ".output_image, .input_image {height: 40rem !important; width: 100% !important;}"
choices = [
    "vi"
]
gr.Interface(
    inference,
    [gr.inputs.Image(type='filepath', label='Input'),gr.inputs.CheckboxGroup(choices, type="value", default=['vi'], label='language')],
    [gr.outputs.Image(type='pil', label='Output'), gr.outputs.Dataframe(type='pandas', headers=['easyOCR','vietOCR', 'confidence'])],
    title=title,
    description=description,
    article=article,
    examples=examples,
    css=css,
    enable_queue=True
    ).launch(debug=True)