GitLab

* reduce 1e-12 to make difference between variables

If 1e-12 is used to make a difference between t1 and t2, then later (t2-t1) might become 0 as 1e-12 can't affect to the number (at least true in pytorch, verified in v1.5.0+).

* Revert "reduce 1e-12 to make difference between variables"

This reverts commit 33ec7215d22be59bf0012dea748f5cacc33a0fb8.

* reduce 1e-12 to make difference between variables

If 1e-12 is used to make a difference between t1 and t2, then later (t2-t1) might become 0 as 1e-12 can't affect to the number (at least true in pytorch, verified in v1.5.0+). Not sure whether 1e-7 or 1e-8 worked. But 1e-6 did.

* visuliaztion typos

* fix to visualization

* fixed lint errors by pre-commit

* upgrade to build_anchor_generator

* fixed  isort message

* Update README.md

* Update README_zh-CN.md

* Update version.py

* Update getting_started.md

* Update getting_started.md

* Update changelog.md

* Remove "recent" in the news

* Remove "recent" in the news

* Fix comments

* Fix regnet pretrained weight loading

* Remove unused file

* update point fusion

* remove LIDAR hardcode

* move get_proj_mat_by_coord_type to utils

* fix lint

* remove todo

* fix lint

* FIX Transpose nested lists without Numpy

* Removed unused Numpy import

* Add header for files

* Add header for files

* Add header for files

* Add header for files

* add s3dis dataset

* fix comments; remove _S3DISDataset if favour of ConcatDataset

* fix comments

* fix comments

* fix link in __init__/py

* remove unused import

* add dataset tests

* add pytest.skip

* fix gf3d cfg name

* fix unit test

* minor fix

* replace all cam_intrinsic

* revert cam_intrinsic in test

* remove rect and fix lint

* replace IndoorPointSample to PointSample

* add comments for PointSample transform

* fix bug when far points > num_sample

* refine format

* add unit test

* minor fix

* minor fix

* use type long long and dynamic memory allocation

* use int64_t instead of long long

* [Enhance] Sampling points based on distance metric

* fix typo

* refine unittest

* refine unittest

* refine details & add unittest & refine configs

* remove __repr__ & rename arg

* fix unittest

* add unitest

* refine unittest

* refine code

* refine code

* refine depth calculation

* refine code

* Update README.md

* Update README_zh-CN.md

* Update getting_started.md

* Update getting_started.md

* Update version.py

* Update changelog.md

* Update changelog.md

* Update changelog.md

* Update changelog.md

* Remove the legacy hack function for monocular dataset browsing

* Remove legacy import from stucture/__init__.py

* Remove legacy hacks for nuScenes visualization

* Add deleted show_bboxes initialization in inference.py

* Rollback deletion of hack function in core

* Adjust the deprecationwarning

* add cfg

* add eps format

* update README

* update model zoo

* add paconv readme

* add metafile

* hack drop_ratio

* use base schedule cfg

* remove paconv cuda

* monor fix

* remove paconv cuda config related

* update paconv benchmark

* update model link

* set gravity center origin

* remove filter transform in 3DSSD cfg

* add PAConv decode head

* add config files

* add paconv's correlation loss

* support reg loss in Segmentor class

* minor fix

* add augmentation to configs

* fix ed7 in cfg

* fix bug in corr loss

* enable syncbn in paconv

* rename to loss_regularization

* rename loss_reg to loss_regularize

* use SyncBN

* change weight kernels to kernel weights

* rename corr_loss to reg_loss

* minor fix

* configs fix IndoorPatchPointSample

* fix grouped points minus center error

* update transform_3d & add configs

* merge master

* fix enlarge_size bug

* refine config

* remove cfg files

* minor fix

* add comments on PAConv's ScoreNet

* refine comments

* update compatibility doc

* remove useless lines in transforms_3d

* rename with_loss_regularization to with_regularization_loss

* revert palette change

* remove xavier init from PAConv's ScoreNet

* Fix rotation and dim hacks between nuscbox and our cambox

* Remove hack in the mono browse, fix cam_intrinsic in the RandomFlip3D

* Apply a more suitable hack transformation (dim & yaw hack)

* Fix incorrect transformation of dim in post-processing

* Remove transformation from global to local yaw in kitti mono dataset

* Add comments for cam_intrinsic modification in the doc and code

* Fix typos and invalid links

* Add compatibility doc for fixing nus hacks

* Update benchmark

* extract RGB images for ScanNet

* update scannet_data_utils with rgb

* fix docs an tools; add use_camera to ScanNetDataset

* fix typos is scannet doc

* fix very rare undefined poses for ScanNet

* update ScanNet dataset for more clear MultiViewPipeline

* update compatibility for ScanNet

* fix typo in compatibility doc

* use mmcv.track_parallel_progress for scannet images

* Fix docstring typos and change formats for api-doc to work

* Test on fixing noqa

* Fix noqa

* Fix noqa

* support 3dssd

* support one-stage method

* for lint

* support two_stage

* Support all methods

* remove init_cfg=[] in configs

* test

* support h3dnet

* fix lint error

* fix isort

* fix code style error

* fix imvotenet bug

* rename init_weight->init_weights

* clean comma

* fix test_apis does not init weights

* support newest mmdet and mmcv

* fix test_heads h3dnet bug

* rm *.swp

* remove the wrong code in build.yml

* fix ssn low map

* modify docs

* modified ssn init_config

* modify params in backbone pointnet2_sa_ssg

* add ssn direction init_cfg

* support segmentor

* add conv a=sqrt(5)

* Convmodule uses kaiming_init

* fix centerpointhead init bug

* add second conv2d init cfg

* add unittest to confirm the input is not be modified

* assert gt_bboxes_3d

* rm .swag

* modify docs mmdet version

* adopt fcosmono3d

* add fcos 3d original init method

* add init cfg

* add init cfg

* modify fcos3d init_weights

* deleted error merge code

* deleted useless code

* remove useless code

* add get_flops.py

* modify SSD to support flops calc

* update docs

* support mono3d models

* fix bev_range problem with points in different coordinate systems

* changed checking instance on points to gt_bboxes_3d to make it compatible with test

* sort imports

* converted rt_mat from ndarray to tensor to make it further compatible

* changed if condition of rt_mat to be consistent with code

* Bump to v0.15.0

* fix typo, specific bugfix, install version

* fix LoadMultiViewImageFromFiles

* compatibility about mmcv version

* unravel multi-view img to list

* add unit test

* support 3dssd

* support one-stage method

* for lint

* support two_stage

* Support all methods

* remove init_cfg=[] in configs

* test

* support h3dnet

* fix lint error

* fix isort

* fix code style error

* fix imvotenet bug

* rename init_weight->init_weights

* clean comma

* fix test_apis does not init weights

* support newest mmdet and mmcv

* fix test_heads h3dnet bug

* rm *.swp

* remove the wrong code in build.yml

* fix ssn low map

* modify docs

* modified ssn init_config

* modify params in backbone pointnet2_sa_ssg

* add ssn direction init_cfg

* support segmentor

* add conv a=sqrt(5)

* Convmodule uses kaiming_init

* fix centerpointhead init bug

* add second conv2d init cfg

* add unittest to confirm the input is not be modified

* assert gt_bboxes_3d

* rm .swag

* modify docs mmdet version

* adopt fcosmono3d

* add fcos 3d original init method

* fix mmseg version

* add init cfg in fcos_mono3d.py

* merge newest master

* remove unused code

* modify focs config due to changes of resnet

* support imvoxelnet pointnet2

* modified the dependencies version

* support decode head

* fix inference bug

* modify the useless init_cfg

* fix multi_modality BC-breaking

* fix error blank

* modify docs error

* group-free-3d head

* GroupFree3DNet->VoteNet

* modify docstring

* bugfix: calculate pts_instance_label, decoder self/cross posembed init

* support GroupFree3DNet, modify scannnet train config

* support point cloud input features dim = 0

* add groupfree3dnet test case

* bugfix: softmax in decode boxes

* support multi-stage predictions

* modify GroupFree3DMultiheadAttention input parameters

* refactor: support sunrgbd-based train

* refactor: support sunrgbd-based train

* fix parts of bug

* modify multi-stage prediction

* fixbug: conv_channels

* bugfix: permute

* bugfix: permute

* bugfix: expand

* fix MAX_NUM_OBJ=64

* 4 gpu training, score_thr = 0.0

* modify config, repeattime=1

* bigfix: expand

* modify: GroupFree3DMHA, build_positional_encoding

* modify: GroupFree3DMHA, build_positional_encoding

* bugfix: torch.nn

* bugfix: mean loss

* residual -> identity

* fix name: DropOut -> Dropout

* delete sunrgbd-based congfig

* Fix: trailing whitespace

* suffix -> prefix

* bugfix: groupfree3d config
Co-authored-by: jinhui <PJLAB\jinhui@shai14001019l.pjlab.org>

* refactor K and Rt to depth2img for SUN RGB-D

* fix lint

* update 3 tests

* fix extra calib key and comments

* remove calib from browse_dataset

* fix cam to depth; rename return_z

* remove mmseg max ver limit

* install mmdet 2.11.0

* mmseg < 0.14.1

* change mmseg to mmsegmentation

* fix mmseg to 0.14.0

* first try of kitti

* fix python3

* imvoxelnet is ready for open-mmlab/mmdetection3d

* apply pre-commit

* update to ConvModule and AlignedAnchor3DGenerator

* add unit tests

* fix torch.Tensor in docstrings

* revert anchor ranges