search

isl-org / Open3D-PointNet2-Semantic3D

Semantic3D segmentation with Open3D and PointNet++
Other
508 stars 113 forks source link

ask for kitti demo #42

Open amiltonwong opened 5 years ago

amiltonwong commented 5 years ago

Hi, @yxlao ,

The provided code only guides how to deal with Semantic3D dataset. From your Open3D 0.5 introduction video, real-time inference in KITTI lidar dataset is presented. Would you also update the pipeline for KITTI lidar dataset?

THX!

yxlao commented 5 years ago

@amiltonwong For training, use train.py and semantic_no_color.json config. For inference, use kitti_predict.py.

XuanAo1016 commented 5 years ago

hi @yxlao I think we also miss the preprocessing part from raw kitti to pcd files

iris0329 commented 5 years ago

Hi @coronax1016, If you just test the pre-trained model provided by the author, you don't have to deal with raw KITTI data, because the python library imported in the 'kitti_predict.py' can handle KITTI data by itself. I only download the KITTI, and run 'kitti_predict.py', it works !

davidlmljll commented 4 years ago

Hello @iris0329, can you tell me how you run the 'kitti_predict.py' and thedata needs to be preprocessed?