search

xingyul / flownet3d

FlowNet3D: Learning Scene Flow in 3D Point Clouds (CVPR 2019)
MIT License
363 stars 83 forks source link

Using only XYZ values #39

Closed Shubhendu-Jena closed 4 years ago

Shubhendu-Jena commented 4 years ago

Hi, thanks a lot for the previous reply. Just another doubt, in the function sample_and_group, if the flag use_xyz is set to true, we have : new_points = tf.concat([grouped_xyz, grouped_points], axis=-1) # (batch_size, npoint, nample, 3+channel). To use XYZ values only, instead of setting (R,G,B) = (0,0,0), I was doing new_points = grouped_xyz which gives me significantly different results. I can't figure out why this creates that much of a difference as I was grouping X,Y,Z coordinates without the concatenated (R,G,B) = (0,0,0) values. Can you please help me with this?

xingyul commented 4 years ago

Note *_points actually denotes the features, either the RGB values or the learned features from the previous layer. The variable name is a little confusing (and sorry about that). So you can only do new_points = grouped_xyz for the first layer. If you do that for the second layer and beyond, the features learned for the that layer is not passed to subsequent layers, so it will fail.

Shubhendu-Jena commented 4 years ago

Thank you!