In the function getitem in flying_things_dataset.py, the point clouds, stored in the cache, are not randomly sampled with 2048 points and are returned with 8192 points. In this case, in the function get_batch in train.py, only first 2048 points of these point clouds will be used.
So I guess, there might exist an indentation problem between line47 and line65 in flying_things_dataset.py.
In the function getitem in flying_things_dataset.py, the point clouds, stored in the cache, are not randomly sampled with 2048 points and are returned with 8192 points. In this case, in the function get_batch in train.py, only first 2048 points of these point clouds will be used.
So I guess, there might exist an indentation problem between line47 and line65 in flying_things_dataset.py.