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

Low performance of motorcyclist and other-ground classes #37

Closed JBKnights closed 3 years ago

JBKnights commented 3 years ago

Hello,

For the pre-trained model included, I noticed that performance on the classes motorcyclist and ground-other are close to zero, while in the paper their reported accuracy is much higher. Is this because of the lack of point refinement module in this repository, or is there some other explanation as to the significant difference in result here?

xinge008 commented 3 years ago

Due to the limited number of points of motorcyclist and other-ground, the results on validation set are very poor, even close to zero.

The results reported in the paper are from test set, which is more balanced compared to validation set.