Can't run on whole dataset

[Adnan1986]

Hi,

I wen to run the code based on pytorch_version instaruction(cpu). When I am going to run the BASS.py for some number of superpixels, There is no warning.errors, while for some number of superpixels, I got following errors for some images:

I alread had some modification:
Nk.index_add_(0, argmax, Global.ones) to Nk.index_add_(0, argmax.long(), Global.ones)

This one fixed a little bit but, still for some number of superpixel I got Segmentation fault (core dumped)

Is there a way to fix this.

Thanks

[uzielroy]

Hi, please provide additional information.
On what image are you trying to run BASS?
What is the initial number of SP?
Please try and run in verbose. (-v)

[Adnan1986]

Hi, please provide additional information. On what image are you trying to run BASS? What is the initial number of SP? Please try and run in verbose. (-v)

Thank you for your tips.
In fact since I was running the code within BSDS dataset and various number of SPs, for sp=600, each time for different image I got Segmentation fault (core dumped). but now by adding verbose(--v), I run the code for entire dataset without problem.
now I got this error again for sp=1400 with the image=201080 and 20069.
It sounds this error would happen randomly.