✨ LaspyReader for reading LAS/LAZ point cloud files

[weiji14]

An iterable-style DataPipe for point cloud data!

I/O handled using laspy.

Preview at https://zen3geo--137.org.readthedocs.build/en/137/api.html#module-zen3geo.datapipes.laspy

Usage:

from torchdata.datapipes.iter import IterableWrapper
from zen3geo.datapipes import LaspyReader

# Read in LAZ data using DataPipe
file_url: str = "https://opentopography.s3.sdsc.edu/pc-bulk/NZ19_Wellington/CL2_BQ31_2019_1000_2138.laz"
dp = IterableWrapper(iterable=[file_url])
_, dp_stream = dp.read_from_http().unzip(sequence_length=2)
dp_laspy = dp_stream.read_from_laspy()

# Loop or iterate over the DataPipe stream
it = iter(dp_laspy)
lasdata = next(it)
lasdata.header
# <LasHeader(1.4, <PointFormat(6, 0 bytes of extra dims)>)>
lasdata.xyz
# array([[ 1.74977156e+06,  5.42749877e+06, -7.24000000e-01],
#        [ 1.74977152e+06,  5.42749846e+06, -7.08000000e-01],
#        [ 1.74977148e+06,  5.42749815e+06, -7.00000000e-01],
#        ...,
#        [ 1.74976026e+06,  5.42756798e+06, -4.42000000e-01],
#        [ 1.74976029e+06,  5.42756829e+06, -4.17000000e-01],
#        [ 1.74976032e+06,  5.42756862e+06, -4.04000000e-01]])

Note that since laspy is made an optional dependency, users would need to do pip install zen3geo[lidar] to install the extra 'lidar' packages that includes laspy[lazrs].

TODO:

• Add laspy as an optional dependency
• Initial implementation of LaspyReaderIterDataPipe
• Add unit tests to read COPC/LAS/LAZ files
• Properly document new lidar extras