Datasets:

Modalities:
Image
Text
Formats:
csv
Languages:
English
ArXiv:
Libraries:
Datasets
Dask
File size: 2,747 Bytes
54d9099
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
# -*- coding: utf-8 -*-
"""

This file contains a list of python script and their purpuse and status

"""
"""

0_environment.yml

This file includes a setup of the python environment requried to run the scripts.

Please note that installation of GPU resources are not added. The model wil 

automatically try to run the script on the GPU (if installed) and will otherwise

use the CPU.



"""

"""

1a_dataset_statistics.py

This is used to compute statistics of the dataset. 

1. Calculates Total Pixel Area (Resolution * Image Count).

2. Calculates "No Defect" (Background) pixel counts.

3. Calculates Pixel Percentages for all categories.

4. Maintains the TA, TB, TC split.

"""

"""

1b_histogram_plot.py

This script reads the segmented masks and plots histograms of the defect size 

distribution. It generates:

1. Individual plots for all datasets and individual plots for the tunnels 

TA, TB, TC.

2. A combined subplot figure comparing TA, TB, and TC.



The user chose which defect that should be plotted.

"""


"""

1c_create_classification.py

this scripts reads the csv files that contain information about images with and

withou cracks. Based on this, three classification datasets are created in the

folder "3_classification", i.e. TA, TB and TC. Each folder contains the 

subfolder "crack" and "no_crack"

"""

"""

2_train_CNN.py



This script trains a UNet segmentaiton model for a single detection class. 

The user defines the "Session_Name" which is the output folder for the saved 

model, plots and metrics. 

The user use the Global Configuration to adjust parameters. This includdes: 

 - A weight factor is included for imbalanced datasets.

- Data used for Traning, Evaluation and Testing is based on csv files.

 - The script creates masks used for the fastai packaage which use 1 for defect

 and 0 for background. The user defines the pixel value for the class they want 

to train the model for. 

 - Model training parameters are easily adjusted.

 - Output includes plots of top 5 best and worst predictions of cracks and 

 txt files with a summary of the metrics

"""

"""

2b_plot_training.py

This script reads the csv output from training and creates a plot of training 

and validation loss in one plot and IoU and F1-score in a second plot. 

User only needs to change "TRAINING_DATA" to correct training set.

"""

"""

3_evaluate_CNN.py

This script loads a pre-trained model and evaluate its performance on a list 

of datasets. The output is a .txt file with metrics. Naming of the file is based on

the SESSION_NAME and metrics for each eavluation is added in the txt file in

sequence, i.e. the metrics for all evaluation using the same model is stored in

the same file.

"""