Example of usage as a python library needed about gt_im_s HOT 3 CLOSED

Yes, because that yaml file contains path to image and init_contour file for that image. You can use it like that too:

import argparse


parser = argparse.ArgumentParser()
parser.add_argument("-i", "--input", help="yaml file with GameTheoreticFramework object", required=True)
parser.add_argument("-in", "--image_name", help="name of the ph2 image (eg. IMD002)",
                    required=True)
parser.add_argument("-pb", "--path_beginning", help="beginning of the image path, added to each path",
                    required=True)
args = parser.parse_args()

with open(args.input) as file:
    gt_segmentation = yaml.load(file, Loader=yaml.Loader)

img_n = args.image_name
path_beg = args.path_beginning

gt_segmentation.image_path = path_beg + r'/%s/%s_Dermoscopic_Image/%s.bmp' % (img_n, img_n, img_n)
gt_segmentation.init_contours = path_beg + r'/init_lesion/%s_lesion.png' % (img_n)

In above example you can use single yaml file for multiple images. You pass an command line argument with some image name, that is used to construct paths to the image itself and to file file with init_contours, then whatever paths were saved in the yaml file get overwritten with this new paths.

This is how an file with init_contours can look like:

It works with any extension supported by openCV function imread
(https://docs.opencv.org/3.4/d4/da8/group__imgcodecs.html#ga288b8b3da0892bd651fce07b3bbd3a56).

from gt_im_s.

1. I made some changes in the game_theoretic_framework class so it works as a library and added an script with example of such usage.
   If you want an image to be stored as a numpy array then set full_init to True when creating the yaml file. I don't recommend it though, because storing images in a yaml file is inefficient.
   Colour images need to be converted to grayscale (it's done automatically when using load_image), because with that kind of images in mind whole algorithm was developed.

Example output:

Yaml file used to acquire that output mask:

!!python/object:game_theoretic_framework.GameTheoreticFramework
b_cost_interlaced: null
clique_size: 4
contours: null
differential_evolution_bounds_width: 0.55
expected_val_in: 1
expected_val_out: 0
image: null
image_gradient: null
image_path: .\obrazki_do_testow\IMD004\IMD004_Dermoscopic_Image\IMD004.bmp
img_gradient_ksize: 27
init_b_cost_interlaced: null
init_contours: .\obrazki_do_testow\IMD004\IMD004_lesion\IMD004_lesion.bmp
init_fourier_coeffs_first_part: null
init_fourier_coeffs_second_part: null
init_img_seg: null
init_tr: null
iter_num: 0
max_iterations: 3
object_brighter_than_background: false
order_of_fourier_coeffs: 14
p2c_acc: 2000
prior_b_cost: 0
region_segmentation: null
scaling_const_alpha: 0.1
scaling_const_beta: 0.1
sm_const: 14

2. In utils there is function hair_removal that removes hair from given image.

from gt_im_s.

Ok, so for multiple images I need to create a yaml file similar to this one and an init_contours image for each one of them? How such an image should look like, @jaroslaw243 ?

Does it work for other extensions than *.bmp?

from gt_im_s.