Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -0,0 +1,151 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import gradio as gr
|
| 2 |
+
import tensorflow as tf
|
| 3 |
+
import keras_ocr
|
| 4 |
+
import requests
|
| 5 |
+
import cv2
|
| 6 |
+
import os
|
| 7 |
+
import csv
|
| 8 |
+
import numpy as np
|
| 9 |
+
import pandas as pd
|
| 10 |
+
import huggingface_hub
|
| 11 |
+
from huggingface_hub import Repository
|
| 12 |
+
from datetime import datetime
|
| 13 |
+
import scipy.ndimage.interpolation as inter
|
| 14 |
+
import easyocr
|
| 15 |
+
import datasets
|
| 16 |
+
from datasets import load_dataset, Image
|
| 17 |
+
from PIL import Image
|
| 18 |
+
from paddleocr import PaddleOCR
|
| 19 |
+
import pytesseract
|
| 20 |
+
from pdf2image import convert_from_path
|
| 21 |
+
from PIL import Image
|
| 22 |
+
import os
|
| 23 |
+
import tempfile
|
| 24 |
+
from openai import OpenAI
|
| 25 |
+
|
| 26 |
+
# Function to perform OCR
|
| 27 |
+
def ocr(input_file, lang='fas'): # 'fas': Persian language (Farsi)
|
| 28 |
+
extracted_text = ""
|
| 29 |
+
# if isinstance(input_file, Image.Image): # If the input is an image
|
| 30 |
+
text = pytesseract.image_to_string(input_file, lang=lang)
|
| 31 |
+
extracted_text = text
|
| 32 |
+
|
| 33 |
+
return extracted_text
|
| 34 |
+
|
| 35 |
+
"""
|
| 36 |
+
Paddle OCR
|
| 37 |
+
"""
|
| 38 |
+
def ocr_with_paddle(img):
|
| 39 |
+
finaltext = ''
|
| 40 |
+
ocr = PaddleOCR(lang='fa', use_angle_cls=True)
|
| 41 |
+
# img_path = 'exp.jpeg'
|
| 42 |
+
result = ocr.ocr(img)
|
| 43 |
+
|
| 44 |
+
for i in range(len(result[0])):
|
| 45 |
+
text = result[0][i][1][0]
|
| 46 |
+
finaltext += ' '+ text
|
| 47 |
+
return finaltext
|
| 48 |
+
|
| 49 |
+
"""
|
| 50 |
+
Keras OCR
|
| 51 |
+
"""
|
| 52 |
+
def ocr_with_keras(img):
|
| 53 |
+
output_text = ''
|
| 54 |
+
pipeline=keras_ocr.pipeline.Pipeline()
|
| 55 |
+
images=[keras_ocr.tools.read(img)]
|
| 56 |
+
predictions=pipeline.recognize(images)
|
| 57 |
+
first=predictions[0]
|
| 58 |
+
for text,box in first:
|
| 59 |
+
output_text += ' '+ text
|
| 60 |
+
return output_text
|
| 61 |
+
|
| 62 |
+
"""
|
| 63 |
+
easy OCR
|
| 64 |
+
"""
|
| 65 |
+
# gray scale image
|
| 66 |
+
def get_grayscale(image):
|
| 67 |
+
return cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
|
| 68 |
+
|
| 69 |
+
# Thresholding or Binarization
|
| 70 |
+
def thresholding(src):
|
| 71 |
+
return cv2.threshold(src,127,255, cv2.THRESH_TOZERO)[1]
|
| 72 |
+
def ocr_with_easy(img):
|
| 73 |
+
gray_scale_image=get_grayscale(img)
|
| 74 |
+
thresholding(gray_scale_image)
|
| 75 |
+
cv2.imwrite('image.png',gray_scale_image)
|
| 76 |
+
reader = easyocr.Reader(['fa','en'])
|
| 77 |
+
bounds = reader.readtext('image.png',paragraph="False",detail = 0)
|
| 78 |
+
bounds = ''.join(bounds)
|
| 79 |
+
return bounds
|
| 80 |
+
|
| 81 |
+
"""
|
| 82 |
+
Generate OCR
|
| 83 |
+
"""
|
| 84 |
+
|
| 85 |
+
def process(input_type, gr_img, lang):
|
| 86 |
+
|
| 87 |
+
if isinstance(gr_img, np.ndarray):
|
| 88 |
+
# Create a temporary file to save the image
|
| 89 |
+
with tempfile.NamedTemporaryFile(suffix=".png", delete=False) as temp_file:
|
| 90 |
+
temp_path = temp_file.name
|
| 91 |
+
# Save the NumPy array as an image file
|
| 92 |
+
Image.fromarray(gr_img).save(temp_path)
|
| 93 |
+
|
| 94 |
+
image = Image.open(temp_path)
|
| 95 |
+
# image = file
|
| 96 |
+
extracted_text = ocr(image, lang)
|
| 97 |
+
return extracted_text
|
| 98 |
+
|
| 99 |
+
|
| 100 |
+
def generate_ocr(Method, img):
|
| 101 |
+
|
| 102 |
+
text_output = ''
|
| 103 |
+
|
| 104 |
+
add_csv = []
|
| 105 |
+
image_id = 1
|
| 106 |
+
if Method == 'Version3':
|
| 107 |
+
text_output = process("img", img, "fas")
|
| 108 |
+
|
| 109 |
+
if not text_output:
|
| 110 |
+
return text_output
|
| 111 |
+
else:
|
| 112 |
+
client = OpenAI(
|
| 113 |
+
base_url="https://openrouter.ai/api/v1",
|
| 114 |
+
api_key="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", # Get from https://openrouter.ai
|
| 115 |
+
)
|
| 116 |
+
|
| 117 |
+
response = client.chat.completions.create(
|
| 118 |
+
model="deepseek/deepseek-r1:free",
|
| 119 |
+
messages=[{"role": "user", "content": f"""متن زیر پر از ایرادات تایپی و نگارشی است اصلاح کن و فقط متن اصلاح شده رو بهم بده
|
| 120 |
+
اطلاعات اظافه رو نیار و فقط متن اصلاح شده رو برگردون
|
| 121 |
+
{text_output}
|
| 122 |
+
"""}]
|
| 123 |
+
)
|
| 124 |
+
|
| 125 |
+
return response.choices[0].message.content
|
| 126 |
+
|
| 127 |
+
|
| 128 |
+
"""
|
| 129 |
+
Create user interface for OCR demo
|
| 130 |
+
"""
|
| 131 |
+
|
| 132 |
+
# image = gr.Image(shape=(300, 300))
|
| 133 |
+
# image = gr.File(label="Upload PDF/Image")
|
| 134 |
+
image = gr.Image()
|
| 135 |
+
method = "Version3"
|
| 136 |
+
output = gr.Textbox(label="Output")
|
| 137 |
+
|
| 138 |
+
demo = gr.Interface(
|
| 139 |
+
generate_ocr,
|
| 140 |
+
[method, image],
|
| 141 |
+
output,
|
| 142 |
+
title="Persian Optical Character Recognition",
|
| 143 |
+
css=".gradio-container {background-color: lightgray} #radio_div {background-color: #FFD8B4; font-size: 40px;}",
|
| 144 |
+
article = """<p style='text-align: center;'>Feel free to give us your thoughts on this demo and please contact us at
|
| 145 |
+
<a href="mailto:zaribar2928@gmail.com" target="_blank">zaribar2928@gmail.com</a>
|
| 146 |
+
<p style='text-align: center;'>Developed by: <a href="https://github.com/mohammad2928" target="_blank">Mohammad Mahmoudi</a></p>"""
|
| 147 |
+
|
| 148 |
+
|
| 149 |
+
)
|
| 150 |
+
# demo.launch(enable_queue = False)
|
| 151 |
+
demo.launch(share=True)
|