search

Na-Z / attMPTI

[CVPR 2021] Few-shot 3D Point Cloud Semantic Segmentation
MIT License
159 stars 29 forks source link

Training Time #20

Open chunbolang opened 2 years ago

chunbolang commented 2 years ago

@Na-Z Hi, thanks for sharing the code!

I am new to 3d point could segmentation. Is this task GPU-friendly under the few-shot setting? How long did it take to run the experiment on each dataset?

Looking forward to your reply. Thanks again~

Best,

Zhenmao1 commented 2 years ago

I test the model on GPU-RTX3070. Pre-train need about 4 hours, train need 8 hours, eval need about 15 min. It is not GPU-friendly, requesting for a large memory, I think at lest need 4G memory（batch 4）

chunbolang commented 2 years ago

Thanks for your instant reply!

It seems that the batch size of the training process (not pre-train) is set to 1. Is this the universal setting for 3D point cloud segmentation?