How many corrupt samples of which effect get created?

[Petros626]

Hey,

When we got for example the KITTI dataset in the required format with 7481 labeled samples, how many corrupted samples(KITTI-C) of each effect like fog, snow, rain etc. will be created? The same amount of samples or each effect is applied once or several times on the data?

Thanks in advance!

[ldkong1205]

Hey,

When we got for example the KITTI dataset in the required format with 7481 labeled samples, how many corrupted samples(KITTI-C) of each effect like fog, snow, rain etc. will be created? The same amount of samples or each effect is applied once or several times on the data?

Thanks in advance!

Hi @Petros626, thanks for your interest in our work!

For each corruption effect, there will be three levels of severity to simulate diverse scenarios. Therefore, the total number of new samples created will be:

7481 x Number of Corruptions x 3

[Petros626]

@ldkong1205 would it be possible to change the code to get the following:

1. Use e.g. 3712 validation samples from the 7481 samples?

3712 x Number of Corruptions x 3 (severity level)

2. Create whole corrupt datasets with just one corruption effect?

7481 or 3712 x First Corruption (e.g. snow) x 3
7481 or 3712 x Second Corruption (e.g. rain) x 3
7481 or 3712 x Third Corruption (e.g. etc.) x 3