Implementation details for EQ-Net

[drprojects]

Hello, thanks for sharing your great code and paper !

I read your Q-Net paper and supplementary material in details, but there was some practical information I could not find. Would you mind sharing the following for your EQ-Net model (the semantic segmentation with SparseConv encoder):

• number of parameters in the encoder / Q-decoder
• number of GPUs used for training / inference
• training (and maybe inference ?) time for S3DIS fold 5

Thanks in advance for those !

[tomztyang]

Hi, thanks for your interest.

• Number of parameters: we have provided a trained EQ-Net on ScanNet. You can simply obtain the number of parameters through that pre-trained weight.
• We use 8 1080 / 2080Ti GPUs for training.
• Hope @llijiang can help you with that.

Thanks, Zetong

[drprojects]

Thank you @tomztyang for your reply ! I will check the parameters myself. Looking forward to @llijiang 's insights on training and inference speeds, then ! Best, Damien