| Commit message (Collapse) | Author | Age |
|---|
| | |
|
| |
|
|
| |
https://build.opensuse.org/package/view_file/science/python-pcl/visualization18.patch
|
| | |
|
| |
|
|
|
|
|
|
|
| |
*Create Sphinx docs(#126#126, v0.3 Release Task list)
*Separate Test Module(#126#126, v0.3 Release Task list)
*visual.Show() Methods default argument Replace binary string(#124)
*SegmentationNormal set_axis Function Implemented(#133)
|
| |
|
|
|
| |
(#137)
This reverts commit 29c136712bcf02d7b831a979f05a5eca6b35bc75.
|
| |
|
| |
This reverts commit 04154d9ddad1cb80c52ef66c95e7ca8890e8fdeb.
|
| |
|
|
|
|
|
| |
Thanks for the PR.
About test code.
0.3 At the time of release, we will implement it at once.
|
| |
|
|
|
|
| |
Update readme.rst
travis CI mac test add
Appveyor pcl 1.8.1 test add
reference count error
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
I deleted the comment by merging mistake, so I modified the code.
I deleted garbled comments.
use size property
__getbuffer__ modified(parameter timing change : _view_count)
Appveyor setting modified(use test.py -> test_pcl.py)
test item add(test_pcl.py modified test_asarray[copy test.py])
* use size property
* __getbuffer__ modified(always _view_count add)
* Appveyor modified(use test.py -> test_pcl.py)
test item add(test_pcl.py modified test_asarray[copy test.py])
__getbuffer__ modified(always _view_count add[other PointCloud pxi])
|
| |
|
|
|
|
|
|
|
|
| |
As a result of review, there was no problem.
Thank you for fixing.
(same pcl\point_types.h PointXYZI define)
Appveyor build errors,
It is because the comment part has been deleted at merging.
We will fix it as soon as possible.
|
| |
|
|
|
|
| |
Build support in Windows environment(test 1.6, not test 1.72/1.8)
Reference example of official website Partial implementation
Adding automatic test configuration file(AppVeyor/TravisCI)
|
| |\
| |
| | |
Fix bug in buffer protocol support
|
| | |
| |
| |
| |
| | |
The view count was not correctly incremented beyond 1, so having two
views and releasing one would bring it back to 0.
|
| | | |
|
| | |
| |
| |
| | |
SACSSegmentationFromNormal for use with cone fitting.
|
| |/
|
|
| |
figure out how to subclass to make it work.
|
| |\
| |
| | |
Getters and setters for StatisticalOutlierRemovalFilter
|
| | | |
|
| |/ |
|
| | |
|
| |
|
|
| |
Left-over from cherry-picking off another branch.
|
| | |
|
| | |
|
| |
|
|
|
|
|
| |
Changes the PointCloud object layout to store a shared_ptr. This
actually simplifies quite a bit of the code (fewer casts), at the
expense of a hack to assign to a shared_ptr and a Cython 0.21
requirement.
|
| | |
|
| |\
| |
| | |
pickle support for pcl.PointCloud
|
| | | |
|
| |\ \
| |/
|/| |
Registration: ICP, GICP, ICP_NL
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| |\ \
| |/
|/|
| |
| | |
Conflicts:
examples/kdtree.py
|
| | | |
|
| | |
| |
| |
| |
| | |
Translates their exceptions to RuntimeError. Not pretty, but better
than abort(). Needs a unit test.
|
| | |
| |
| |
| | |
All of these can throw C++ exceptions when running out of memory.
|
| | | |
|
| | | |
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Body of _nearest_k is entirely white in cython -a output.
Also a tad more type-safety with the "not None" clause.
Performance:
>>> import numpy as np
>>> import pcl
>>> rng = np.random.RandomState(42)
>>> p1 = pcl.PointCloud(rng.randn(100000, 3).astype(np.float32))
>>> p2 = pcl.PointCloud(rng.randn(4000, 3).astype(np.float32))
>>> kd = pcl.KdTreeFLANN(p1)
>>> %timeit kd.nearest_k_search_for_cloud(p2)
100 loops, best of 3: 14.2 ms per loop
>>> %timeit kd.nearest_k_search_for_cloud(p2, 3)
100 loops, best of 3: 15.4 ms per loop
>>> %timeit kd.nearest_k_search_for_cloud(p2, 30)
10 loops, best of 3: 28.3 ms per loop
Previously:
>>> %timeit kd.nearest_k_search_for_cloud(p2)
10 loops, best of 3: 42.9 ms per loop
>>> %timeit kd.nearest_k_search_for_cloud(p2, 3)
10 loops, best of 3: 43.5 ms per loop
>>> %timeit kd.nearest_k_search_for_cloud(p2, 30)
10 loops, best of 3: 57.5 ms per loop
|
| |/
|
|
|
| |
Previously, constructing a KdTreeFLANN object made no sense since there
was no way to set its input cloud. It would segfault at first use.
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
This is the segfault reported in #28.
__cinit__ unconditionally calls its base class version, so two objects
got allocated and one of them was leaked. Fixed by moving the actual
allocation to __init__.
Similarly, __dealloc__ would call its base class version and a "double
delete" would follow. Fixed by removing the child class's __dealloc__,
and set self.me to NULL explicitly in OctreePointCloud.__dealloc__ for
added safety.
|
| |
|
|
| |
Used to trigger a C assert/abort in PCL 1.7.1.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Uses a hack to get around limitations in Cython, see pcl/indexing.hpp.
See https://groups.google.com/forum/#!topic/cython-users/AAo7MQFLZyw
for explanation.
Also, we no longer need to use the at member when bounds checking is not
needed:
>>> import pcl
>>> p = pcl.PointCloud()
>>> %timeit p.from_array(a)
100000 loops, best of 3: 17.5 µs per loop
This used to be 23.4 µs per loop.
|
| | |
|
| | |
|
| | |
|
| | |
|
|
|
Allows splitting the big pcl.pyx module and writing parts
in pure Python to speed up compilation.
|
