search

Tsinghua-MARS-Lab / DenseTNT

MIT License
461 stars 121 forks source link

Question about goals sampling #15

Closed ultimatedigiman closed 2 years ago

ultimatedigiman commented 2 years ago

Hi,

In the ICCV paper,I noticed that you sampled dense goals along the road, as illustrated in your paper:

image

but in this code repo, you sampled goals along the lane centerline, which is as same as the vanilla TNT: https://github.com/Tsinghua-MARS-Lab/DenseTNT/blob/398b8ceb5eb741d134c3e96a7515afef764d6a2f/src/dataset_argoverse.py#L125-L137

The parameter include_beside of function get_subdivide_points() will produce a denser goal set, but it seems you are not using it.

GentleSmile commented 2 years ago

This is a trick to reduce the number of sampled goals. The dense goal sampling is actually performed at https://github.com/Tsinghua-MARS-Lab/DenseTNT/blob/398b8ceb5eb741d134c3e96a7515afef764d6a2f/src/modeling/decoder.py#L144.

ultimatedigiman commented 2 years ago

Thanks! What does the comment "not compatible argo" in the sampling function get_neighbour_points() mean? https://github.com/Tsinghua-MARS-Lab/DenseTNT/blob/398b8ceb5eb741d134c3e96a7515afef764d6a2f/src/utils.py#L1226-L1229