Created
July 17, 2018 16:29
-
-
Save ghandic/7f9771d5e03a738d8ddddcd9ec808b42 to your computer and use it in GitHub Desktop.
Running Tesseract with lots of small images is much slower than multipage TIFF, so make a multipage tiff (to be implemented with multiprocessing for the major speed gains)
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
## Tesseract4 | |
# 2 images into tiff: 843 msec 796 | |
# 10 images into tiff: 1003 msec | |
# 53 images (same as page 1) into tiff: 1740 msec | |
## Tesseract3 | |
# 2 images into tiff: 727 msec | |
# 10 images into tiff: 816 msec | |
# 53 images (same as page 1) into tiff: 1620 msec | |
import tempfile | |
from PIL import Image | |
import cv2 | |
import subprocess | |
import pandas as pd | |
from io import StringIO, BytesIO | |
import os | |
import requests | |
from collections import namedtuple | |
TiffPage = namedtuple('TiffPage', 'field image') | |
TESS_EXE='tesseract4' | |
def tess_from_array(arr): | |
tiff_res = {tp.field:'' for tp in arr} | |
pages = {str(i+1): arr[i] for i in range(len(arr))} | |
tiff_list = [tp.image for tp in pages.values()] | |
tmp = tempfile.TemporaryFile() | |
tiff_list[0].save(tmp, 'TIFF', save_all=True, append_images=tiff_list[1:], dpi=(300,300)) | |
tmp.seek(0) | |
tess_process = subprocess.Popen('{} - stdout --psm 7 -c tessedit_unrej_any_wd=1 -c load_system_dawg=F -c load_freq_dawg=F --tessdata-dir /usr/local/share/tessdata tsv quiet'.format(TESS_EXE).split(' '), | |
stdin=tmp, stdout=subprocess.PIPE) | |
results= tess_process.communicate() | |
df = pd.read_csv(StringIO(results[0].decode('utf-8').replace('"', '')), sep="\t") | |
df.fillna('', inplace=True) | |
for n in set(df['page_num']): | |
img_res = ' '.join(df[df['page_num'] == n]['text']).strip() | |
tiff_res[pages[str(n)].field] = img_res | |
return tiff_res | |
if __name__ == '__main__': | |
# Image.open(io.BytesIO(requests.get(url).content)) | |
urls = ['http://www.edocfile.com/images/screenfortextsearch.jpg', | |
'https://encrypted-tbn0.gstatic.com/images?q=tbn:ANd9GcTwxoaykiHAb0-LjKgJ15wRlhn5S24CKUcknQdQPWelyGdfsmHp'] | |
arr = [TiffPage(url, Image.open(BytesIO(requests.get(url).content))) for url in urls] | |
print(tess_from_array(arr)) | |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment