icdar-tools 0.0.3

a pip install icdar_tools

These tools are to provide effort by researchers in creating their own working environment This is about dealing with {ICDAR} data It provides you with initial processing tools for training and testing data. It provides tools for calculating the text area using polygon of shapely. Save results from images and text locations as a prelude to calculating precision. And some other tools we will try to "more examples to explain the use later."

These tools have been quoted and written by the {EAST}. Where you can see the original files here. https://github.com/argman/EAST/

These tools depend on several libraries you must provide before use. Like:

-opencv-3.x.x
-numpy
-scipy
-matplotlib
-shapely

use Modules!

import icdar_tools

or

from icdar_tools import icdar
from icdar_tools import icd_util
from icdar_tools import locality_aware_nms
from icdar_tools import data_util

• icdar.py

This module is very important as it is found to serve your time instead of betting a lot of effort and time in order to produce already existing tools, in order to handle the data. Here you will find everything you need, from the future ICDAR Data Processing

From loading the data and locating the texts inside the images and some other things. The following are examples of usage.

1:get_batch()

get_batch(num_workers, **kwargs)

The function works to get the coordinates of the text in the images Through text files with them in the same path It then returns those geometrical coordinates, image names, and images derived from the training images specified by the place of the text only.

use:

data_generator = icdar.get_batch(num_workers=num_readers,
                                         training_data_path='path/to_data/icdar15/train/'
                                         input_size=input_size,
                                         batch_size=batch_size_per_gpu * len(gpus))

reutrn

yield images, image_fns, score_maps, geo_maps, training_masks

2:load_annoataion()

text_polys, text_tags = icdar.load_annoataion(txt_file-name)

3:restore_rectangle_rbox()

text_box_restored = icdar.restore_rectangle_rbox(origin, geometry)

:

• icd_util.py

1 - get_images() The input path should be images

images_list_fullName = icd_util.get_images(path/data/images/)

Repetition is a list of all images in the input path

2 -resize_image()

im_resized, (ratio_h, ratio_w) = icd_util.resize_image(image)

'''
resize image to a size multiple of 32 which is required by the network
:param im: the resized image
:param max_side_len: limit of max image size to avoid out of memory in gpu
:return: the resized image and the resize ratio
'''

• The default setting of the function

  icd_util.resize_image(image, max_side_len=2400)
  

3 - detect()
Here is the conclusion of the model represented in the geometrical map of coordinates and score

Use the threshold to filter the results that look false The borders of the text boxes are then redrawn

return of these boxes and the time of implementation of this processe.

boxes, timer = icd_util.detect(score_map=score, geo_map=geometry, timer=timer)

'''
restore text boxes from score map and geo map
:param score_map:
:param geo_map:
:param timer:
:param score_map_thresh: threshhold for score map
:param box_thresh: threshhold for boxes
:param nms_thres: threshold for nms
:return: boxes and time out
'''

- The default setting of the function

icd_util.detect(score_map, geo_map, timer, score_map_thresh=0.8, box_thresh=0.1, nms_thres=0.2):

• write_result()
  This function gets the image and its name
  The file name is written as the text location in the image

You get the text boxes that are expected for that image
writeing text locations in text files
drawing squares around those texts in the picture
See the font size of the box and font color through passes

color, thickness

Finally a place will be written those 'output_path/'

Images and text files are written into a single folder.

   icd_util.write_result(img ,boxes ,output_dir ,res_file ,img_fn)

• The default setting of the function

  icd_util.write_result(img ,boxes ,output_dir ,res_file ,img_fn ,color=(255, 255, 0),thickness=1, skip = True)

...