|
a |
|
b/code/utils.py |
|
|
1 |
""" |
|
|
2 |
DeepSlide |
|
|
3 |
General helper methods used in other functions. |
|
|
4 |
|
|
|
5 |
Authors: Jason Wei, Behnaz Abdollahi, Saeed Hassanpour |
|
|
6 |
""" |
|
|
7 |
|
|
|
8 |
import datetime |
|
|
9 |
from pathlib import Path |
|
|
10 |
from typing import (Dict, List) |
|
|
11 |
|
|
|
12 |
# Valid image extensions. |
|
|
13 |
IMAGE_EXTS = [".jpg", ".jpeg", ".png", ".svs", ".tif", ".tiff"] |
|
|
14 |
|
|
|
15 |
|
|
|
16 |
def get_classes(folder: Path) -> List[str]: |
|
|
17 |
""" |
|
|
18 |
Find the classes for classification. |
|
|
19 |
|
|
|
20 |
Args: |
|
|
21 |
folder: Folder containing the subfolders named by class. |
|
|
22 |
|
|
|
23 |
Returns: |
|
|
24 |
A list of strings corresponding to the class names. |
|
|
25 |
""" |
|
|
26 |
return sorted([f.name for f in folder.iterdir() if |
|
|
27 |
((folder.joinpath(f.name).is_dir()) and (".DS_Store" not in f.name))], key=str) |
|
|
28 |
|
|
|
29 |
|
|
|
30 |
def get_log_csv_name(log_folder: Path) -> Path: |
|
|
31 |
""" |
|
|
32 |
Find the name of the CSV file for logging. |
|
|
33 |
|
|
|
34 |
Args: |
|
|
35 |
log_folder: Folder to save logging CSV file in. |
|
|
36 |
|
|
|
37 |
Returns: |
|
|
38 |
The path including the filename of the logging CSV file with date information. |
|
|
39 |
""" |
|
|
40 |
now = datetime.datetime.now() |
|
|
41 |
|
|
|
42 |
return log_folder.joinpath(f"log_{now.month}{now.day}{now.year}" |
|
|
43 |
f"_{now.hour}{now.minute}{now.second}.csv") |
|
|
44 |
|
|
|
45 |
|
|
|
46 |
def get_image_names(folder: Path) -> List[Path]: |
|
|
47 |
""" |
|
|
48 |
Find the names and paths of all of the images in a folder. |
|
|
49 |
|
|
|
50 |
Args: |
|
|
51 |
folder: Folder containing images (assume folder only contains images). |
|
|
52 |
|
|
|
53 |
Returns: |
|
|
54 |
A list of the names with paths of the images in a folder. |
|
|
55 |
""" |
|
|
56 |
return sorted([Path(f.name) for f in folder.iterdir() if |
|
|
57 |
((folder.joinpath(f.name).is_file()) and (".DS_Store" not in f.name) and (f.suffix.casefold() in IMAGE_EXTS))], key=str) |
|
|
58 |
|
|
|
59 |
|
|
|
60 |
def get_image_paths(folder: Path) -> List[Path]: |
|
|
61 |
""" |
|
|
62 |
Find the full paths of the images in a folder. |
|
|
63 |
|
|
|
64 |
Args: |
|
|
65 |
folder: Folder containing images (assume folder only contains images). |
|
|
66 |
|
|
|
67 |
Returns: |
|
|
68 |
A list of the full paths to the images in the folder. |
|
|
69 |
""" |
|
|
70 |
return sorted([folder.joinpath(f.name) for f in folder.iterdir() if |
|
|
71 |
((folder.joinpath(f.name).is_file()) and (".DS_Store" not in f.name) and (f.suffix.casefold() in IMAGE_EXTS))], key=str) |
|
|
72 |
|
|
|
73 |
|
|
|
74 |
def get_subfolder_paths(folder: Path) -> List[Path]: |
|
|
75 |
""" |
|
|
76 |
Find the paths of subfolders. |
|
|
77 |
|
|
|
78 |
Args: |
|
|
79 |
folder: Folder to look for subfolders in. |
|
|
80 |
|
|
|
81 |
Returns: |
|
|
82 |
A list containing the paths of the subfolders. |
|
|
83 |
""" |
|
|
84 |
return sorted([folder.joinpath(f.name) for f in folder.iterdir() if |
|
|
85 |
((folder.joinpath(f.name).is_dir()) and (".DS_Store" not in f.name))], key=str) |
|
|
86 |
|
|
|
87 |
|
|
|
88 |
def get_all_image_paths(master_folder: Path) -> List[Path]: |
|
|
89 |
""" |
|
|
90 |
Finds all image paths in subfolders. |
|
|
91 |
|
|
|
92 |
Args: |
|
|
93 |
master_folder: Root folder containing subfolders. |
|
|
94 |
|
|
|
95 |
Returns: |
|
|
96 |
A list of the paths to the images found in the folder. |
|
|
97 |
""" |
|
|
98 |
all_paths = [] |
|
|
99 |
subfolders = get_subfolder_paths(folder=master_folder) |
|
|
100 |
if len(subfolders) > 1: |
|
|
101 |
for subfolder in subfolders: |
|
|
102 |
all_paths += get_image_paths(folder=subfolder) |
|
|
103 |
else: |
|
|
104 |
all_paths = get_image_paths(folder=master_folder) |
|
|
105 |
return all_paths |
|
|
106 |
|
|
|
107 |
|
|
|
108 |
def get_csv_paths(folder: Path) -> List[Path]: |
|
|
109 |
""" |
|
|
110 |
Find the CSV files contained in a folder. |
|
|
111 |
|
|
|
112 |
Args: |
|
|
113 |
folder: Folder to search for CSV files. |
|
|
114 |
|
|
|
115 |
Returns: |
|
|
116 |
A list of the paths to the CSV files in the folder. |
|
|
117 |
""" |
|
|
118 |
return sorted([folder.joinpath(f.name) for f in folder.iterdir() if ( |
|
|
119 |
(folder.joinpath(f.name).is_file()) and ("csv" in f.name) and (".DS_Store" not in f.name))], |
|
|
120 |
key=str) |
|
|
121 |
|
|
|
122 |
|
|
|
123 |
def create_labels(csv_path: Path) -> Dict[str, str]: |
|
|
124 |
""" |
|
|
125 |
Read the labels from a CSV file. |
|
|
126 |
|
|
|
127 |
Args: |
|
|
128 |
csv_path: Path to the CSV file. |
|
|
129 |
|
|
|
130 |
Returns: |
|
|
131 |
A dictionary mapping string filenames to string labels. |
|
|
132 |
""" |
|
|
133 |
with csv_path.open(mode="r") as lines_open: |
|
|
134 |
lines = lines_open.readlines()[1:] |
|
|
135 |
|
|
|
136 |
file_to_gt_label = {} |
|
|
137 |
|
|
|
138 |
for line in lines: |
|
|
139 |
if len(line) > 3: |
|
|
140 |
pieces = line[:-1].split(",") |
|
|
141 |
file = pieces[0] |
|
|
142 |
gt_label = pieces[1] |
|
|
143 |
file_to_gt_label[file] = gt_label |
|
|
144 |
|
|
|
145 |
return file_to_gt_label |