edwardzhou130 / PolarSeg

Implementation for PolarNet: An Improved Grid Representation for Online LiDAR Point Clouds Semantic Segmentation (CVPR 2020)
BSD 3-Clause "New" or "Revised" License
369 stars 80 forks source link

Speed of code #23

Closed rruixxu closed 3 years ago

rruixxu commented 3 years ago

Why is your code so disgusting? The training only supports single GPU, which is too slow. The inference takes time longer than 1 second on GTX 2080Ti, which is way conflict with what you state in paper.

YangZhang4065 commented 3 years ago

abhigoku10 independently verified our 60ms latency with a 1080Ti in issue https://github.com/edwardzhou130/PolarSeg/issues/8 . You need to check your Pytorch GPU setup. We would love to hear any constructive advice. And we are more than happy to help if anyone has any problems. However, toxicity is not welcomed here.