Question about labels

[cbiras]

A label file such as \train\scene-0001\000_occ.npy contains a np array with a shape of (39068,2). I have a few questions about this array. 1) Considering that the voxel space is (200,200,16), the 39068 points from the array represent occupied voxels from the 640000 total voxels? 2) I want to train a network that will use labels of shape (200,200,16), is there a place in the code where you translate the (39068,2) labels to the full voxel space? 3) Is there a way to extract only the voxels visible to the FRONT camera, in case I want to predict only the front scene, and not the whole surround scene? Thank you!

[tongwwt]

1. Yes
2. Please refere to the visualization code, in which the discrtete labels can be transformed into the representation of voxel space.
3. You need to process the data and project the voxel into the FRONT camera and keep the valid voxel for your setting.

[cbiras]

Thank you for your kind response. I do have a few more questions: I see that the flow files contain numpy arrays with shape for eg. (4691,2). 1) The 2 values for each row represent the velocity on both x and y axes? 2) Is there a way to link the flow with the voxel it represents? Is there a way to find out the flow for a specific voxel, from the ground truth provided? Thank you!