fault injecting tool

The tool generate different RGB image from each image sample and saved it in a specific path defined by the user.

To-Do:

-Add snow effect that covers the ground.
-Add different implementation to rain without overlaying.
-Add different implementation to fog without overlaying.
-Add blur area instead of blurring the whole image.
-Add shadow on the ground fault.
-Create better overlay fault samples.
-Simplifying the code in later stage for more understanding.

Description to the files

Fault.py

contain the fault class and the implementation of the faults methods.

main.py

contain a simple graphical user interface for easy use.

Lena.jpg

it is an RGB image used as test sample.

output

the tool create the output folder if it does not exist in the specified path and saves the generated samples in it in a different folder depending on the selected fault.

split_data.py

used to split dataset to (train, test, and val) with specified ratio.

Graphical user interface

Test sample

Result

the faults shown are blur, crack, speckle noise and dead pixels for the strength from 1 to 5

Contact data

Omar Mohammed | University of Siegen