search

charlesq34 / pointnet

PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation
Other
4.73k stars 1.45k forks source link

Own dataset training on Semantic segmenatation. #112

Open PRAhit opened 6 years ago

PRAhit commented 6 years ago

Dear @charlesq34 Thanks for providing the open source for such amazing work. For semantic segmentation, I have converted my raw point cloud data from .bag -> pcd -> .ply -> .hdf5. I have converted it via third party code without using the gen_indoor3d_h5.py. Now when i am doing training, it throws such error. Can you please help me to sought this out.

screenshot from 2018-06-10 20-13-28

luoxiaoliaolan commented 6 years ago

@PRAhit In gen_indoor3d_h5.py, there are two field：“data” and “label”, maybe the third party code don't contain them.

weiweimanger commented 5 years ago

hello，Im sorry to bother you,I d like to know whether you solved the problem,and is it necessary to .bag - > pcd - > .ply - > .hdf5? Does it work if i convert my data in pcd - > .hdf5? hope your reply

arunumd commented 4 years ago

hello，Im sorry to bother you,I d like to know whether you solved the problem,and is it necessary to .bag - > pcd - > .ply - > .hdf5? Does it work if i convert my data in pcd - > .hdf5? hope your reply

Even I am curious to know if it would be possible to directly convert from a .pcd file to a .h5 file. The process of going through pcd -> ply -> h5 seems longer than pcd -> h5 for getting realtime performance !

jediofgever commented 4 years ago

@arunumd have you find a solution ?