search

xinge008 / Cylinder3D

Rank 1st in the leaderboard of SemanticKITTI semantic segmentation (both single-scan and multi-scan) (Nov. 2020) (CVPR2021 Oral)
Apache License 2.0
859 stars 180 forks source link

'std::bad_alloc' error while trying the demo #105

Open acaglayan opened 2 years ago

acaglayan commented 2 years ago

Hi, while trying the below command:

python demo_folder.py --demo-folder demo_folder --save-folder demo_save

I had the following error:

terminate called after throwing an instance of 'std::bad_alloc' what(): std::bad_alloc Aborted

greatly appreciate any idea to fix the problem. thanks.

xinge008 commented 2 years ago

Can you show more details about your environment seting? such as spconv, pytorch, etc;

From the error, it seems a mistake for the installation of spconv.

acaglayan commented 2 years ago

Hi, I'm using the below packages on a debian 10.11 system:

cudatoolkit 11.3.1 cumm-cu113 0.2.6 cython 0.29.24
numpy 1.21.2
pillow 8.4.0
python 3.7.11
python-dateutil 2.8.2
pytorch 1.10.0
pytorch-mutex 1.0
spconv-cu113 2.1.19
torch-scatter 2.0.9
torchaudio 0.10.0
torchvision 0.11.1
yaml 0.2.5

I'm trying to use the trained model on a custom dataset. I think my error is related to that. I couldn't find info on how to apply a trained model to a custom data. So I tried to modify config files for the new data but couldn't make it run. thank you again.