Spaces:
Runtime error
Runtime error
Update files/functions.py
Browse files- files/functions.py +0 -31
files/functions.py
CHANGED
|
@@ -26,37 +26,6 @@ from matplotlib import font_manager
|
|
| 26 |
from PIL import Image, ImageDraw, ImageFont
|
| 27 |
import cv2
|
| 28 |
|
| 29 |
-
import cv2
|
| 30 |
-
import pytesseract
|
| 31 |
-
import numpy as np
|
| 32 |
-
import pandas as pd
|
| 33 |
-
import random
|
| 34 |
-
from langdetect import detect_langs
|
| 35 |
-
from pdf2image import convert_from_path
|
| 36 |
-
from pdf2image.exceptions import PdfReadError
|
| 37 |
-
from pdfminer.high_level import extract_text
|
| 38 |
-
from pdfminer.layout import LTTextContainer
|
| 39 |
-
from pytesseract import Output
|
| 40 |
-
from langdetect import detect_langs
|
| 41 |
-
from transformers import pipeline, AutoModelForTokenClassification, AutoTokenizer
|
| 42 |
-
from collections import defaultdict
|
| 43 |
-
import torch
|
| 44 |
-
from torch import nn
|
| 45 |
-
from torch.nn import functional as F
|
| 46 |
-
from torch.utils.data import Dataset
|
| 47 |
-
from torchvision import models
|
| 48 |
-
import json
|
| 49 |
-
from itertools import groupby
|
| 50 |
-
from operator import itemgetter
|
| 51 |
-
import os
|
| 52 |
-
import re
|
| 53 |
-
import tempfile
|
| 54 |
-
import font_manager
|
| 55 |
-
from pdf2image.exceptions import (
|
| 56 |
-
PDFInfoNotInstalledError,
|
| 57 |
-
PDFPageCountError,
|
| 58 |
-
PDFSyntaxError
|
| 59 |
-
)
|
| 60 |
|
| 61 |
|
| 62 |
# Tesseract
|
|
|
|
| 26 |
from PIL import Image, ImageDraw, ImageFont
|
| 27 |
import cv2
|
| 28 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 29 |
|
| 30 |
|
| 31 |
# Tesseract
|