GitLab

* fix comment errors

* add eval_pipeline in mono-nuscene cfg

* add vis function in nuscene-mono dataset

* refactor vis function to support all three mode boxes proj to img

* add unit test for nuscene-mono show func

* browse_dataset support nuScenes_mono

* add show_results() to SingleStageMono3DDetector

* support mono-3d dataset browse

* support nus_mono_dataset and single_stage_mono_detector show function

* update useful_tools.md docs

* support mono-3d demo

* add unit test

* update demo docs

* fix typos & remove unused comments

* polish docs

* fix if condition to enable KITTI detectors eval on nus/lyft datasets

* fix comments

* compress kitti unit test imgs

* add unit test for inference_multi_modality_detector

* fix typos

* rename init_detector to init_model

* show_result_meshlab support seg task

* add unit test for seg show_result_meshlab

* support inference_segmentor

* support pc seg demo

* add docs

* minor fix

* change function name

* compress demo data size

* update docs

* [Fix] Updating mmcv_version to be consistent with mmseg and to support MMCLI

* updating get_started.md

* updating mminstall to be the pip requirement format

* add configs for pn2 on s3dis

* fix bugs

* add README for PN2 segmentor

* update configs file after experiments

* update README

* add base schedule files for seg configs

* rename pn2_seg config files

* update pn2 links in README

* add metafile for pn2

* build BaseSegmentor for point sem seg

* add encoder-decoder segmentor

* update mmseg dependency

* fix linting errors

* warp predicted seg_mask in dict

* add unit test

* use build_model to wrap detector and segmentor

* fix compatibility with mmseg

* faster sliding inference

* merge master

* configs for training on ScanNet

* fix CI errors

* add comments & fix typos

* hard-code class_weight into configs

* fix logger bugs

* update segmentor unit test

* logger use mmdet3d

* use eps to replace hard-coded 1e-3

* add comments

* replace np operation with torch code

* add comments for class_weight

* add comment for BaseSegmentor.simple_test

* rewrite EncoderDecoder3D to avoid inheriting from mmseg

* fix error direction

* fix aos bug in get_bboxes

* support PN2 decode head

* add mmseg dependency in github workflow

* complete PN2 decode head

* modify backbone pn2 to support seg task & its unit test

* add unit test for PN2 decode_head

* Bump to v0.13.0

* modify version.py and be more specific in changelog.md

* space before brackets

* specific pytorch 1.8 issue

* version in getting_started.md

* fix bugs

* add few bugfix

* Update changelog.md

* Update changelog.md

* Update getting_started.md

* Adjust the orders of bug fixes

* Update getting_started.md

* Update changelog.md
Co-authored-by: twang <30491025+Tai-Wang@users.noreply.github.com>

* fix small bug in get_loading_pipeline()

* adopt eval_pipeline & support seg visualization

* support multi-modality vis

* fix small bugs

* add multi-modality args & support S3DIS vis

* update docs with imgs

* fix typos

* remove registry.py in datasets/

* remove registry.py in ops/pointnet_modules/

* remove registry.py in models/

* refactor builder using mmcv.MODELS registry

* update mmcv min version requirement

* change min mmcv version requirement in mmdet3d/__init__.py

* min mmcv version to 1.3.2

* fix small errors in seg dataset cfgs

* add box field to 3d seg dataset

* fix small bug in S3DIS data generation scripts

* fix small bugs in scannet data generation

* efficient mapping by matrix indexing

* add pointsegmapping unit test

* add max_cat_id args everywhere PointSegClassMapping is called in codebase

* add default value for max_cat_id for safety

* add assertion of max_cat_id

* Support base mono3d dense head and anchor free mono3d head

* Support FCOS3D head

* Support FCOS3D baseline on nuScenes

* Fix an import error caused by update of mmcv/mmdet

* Change img_scale to scale_factor in the MultiScaleFlipAug in the config

* Add pred_bbox2d in the params of anchor_free_mono3d_head

* Add unit test for fcos3d head

* Fix a minor bug when setting img_metas in the unit test

* Add unit test for fcos3d detector

* Simplify the logic of weights initialization

* Add comments to specify the reason of cloning features

* Update head config

* support S3DIS data download and pre-processing (to ScanNet format)

* add S3DIS data for unittest

* add S3DIS semseg dataset class and unittest

* add config file for S3DIS dataset

* add eval_pipeline to S3DIS dataset config file

* clean code for S3DIS pre-processing scripts

* reformat code

* fix small bugs

* resolve conflicts & modify show() to use pipeline

* fix small errors

* polish data pre-processing code

* add more comments about S3DIS dataset

* fix markdown lint error

* Clamp 2D Projection of 3D Votes to Image Boundaries

* 'lint'

* 'lint;

* lint
Co-authored-by: Yezhen Cong <52420115+THU17cyz@users.noreply.github.com>

* reuse pipeline in scannet-det dataset

* reuse pipeline in kitti dataset

* reuse pipeline in lyft dataset

* reuse pipeline in sunrgbd dataset

* reuse pipeline in nuscenes dataset

* reuse pipeline in waymo dataset

* reuse pipeline in scannet-seg dataset

* add eval_pipeline in configs which have custom data pipelines

* move data loading via pipeline to dataset._extract_data() for clarity

* use eval_pipeline in tools/misc/visualize_results.py

* get_pipeline from self when no pipeline is provided

* fix small bugs

* fix small bugs

* simplify and clear code

* remove unnecessary eval_pipeline added

* add comment about why we set self.test_mode=False

* small fix

* modify docs about config

* modify PointShuffle and add unittest

* modify PointsRangeFilter and add unittest

* fix small bugs in IndoorPointSample

* fix small typos in docstring

* Support single stage mono3d detector

* Add single stage mono3d detector in the init

* Support FCOS Mono3D detector

* Minor adjustment of docstring

* Fix detector import

* Move numpy import to header

* Support nuscenes mono3d json info generation

* Support nuscenes mono3d dataset class

* Support attribute and bbox2d prediction in bbox3dnms and bbox3d2result

* Rename dataset class and add comments to 'attrs'

* Support mono3d related pipelines

* Fix unittest for loading 3D annotations

* Add unit test for nuscenes mono3d dataset

* Rename the sample result file

* Upload sample data for mono3d unit test

* Upload sample data for mono3d unit test

* Upload sample image for unit test

* Delete tests/data/nuscenes/samples/LIDAR_TOP/CAM_BACK_LEFT directory

* Add files via upload

* Remove unnecessary 'f'

* Remove unnecessary \ in arguments

* Remove check for pycocotools version because it has been done in the cocodataset

* Remove unnecessary comma, add TODO and change init of attrs in format_results

* Merge RandomFlip3D and RandomFlipMono3D

* Add pytest to check whether cuda is available in the unit test

* Support monocular 3D detection on KITTI dataset

* Add visualization TODO

* Merge nus_mono3d and update dataset init

* Remove duplicated loading images in mono3d

* Remove aos evaluation of bbox2d predictions on KITTI

* Add unit test for kitti mono3d dataset

* Add accidentally deleted classes in the dataset init

* Replace .format with f-string in kitti dataset

* Clean comma

* Toy data for unit test of kitti mono dataset

* Sample image data for unit test

* add format_results function for ScanNet test set results submission

* support ScanNet test set data pre-processing

* replace is_train with test_mode for better consistency

* fix link error in docstring

* uncomplete

* support lidar2img loading

* add box_type_3d args in MVX-Net config file

* support multi-modality demo for LiDAR point clouds

* support multi-modality demo for indoor (depth) point clouds

* move demo data into folder and modify docs

* add input check & more general filename matching

* update docs for demo and add README file for demo

* add score_threshold option to demos

* add data for ScanNet & KITTI dataset multi-modality test

* add multi-modality visualization in ScanNet and KITTI dataset

* add unittest for modified visualization function

* delete saved temp file and dirs in unittests using TemporaryDirectory

* fix typos in docs & move README of demos to docs/

* add demo docs to documentation

* fix link error
Co-authored-by: wHao-Wu <wenhaowu.chn@gmail.com>

* browse_dataset

* move visualization config from 'configs/' to 'configs/_base_'

* refine

* refine

* adding argument option & browsing testing split only

* support get_loading_pipeline & browsing training split

* support get_loading_pipeline & browsing training split

* support get_loading_pipeline & browsing training split

* add condition of RepeatDataset & support the usage of cfg-options & refine useful_tools.md

* add condition of RepeatDataset & support the usage of cfg-options & refine useful_tools.md

* enable dataset browsing with empty gt & update docs/useful_tools

* support all 3d dataset

* fix small typos

* fix small bugs
Co-authored-by: Wuziyi616 <dazitu616@gmail.com>

* Support nuscenes mono3d json info generation

* Support nuscenes mono3d dataset class

* Support attribute and bbox2d prediction in bbox3dnms and bbox3d2result

* Rename dataset class and add comments to 'attrs'

* Support mono3d related pipelines

* Fix unittest for loading 3D annotations

* Add unit test for nuscenes mono3d dataset

* Rename the sample result file

* Upload sample data for mono3d unit test

* Upload sample data for mono3d unit test

* Upload sample image for unit test

* Delete tests/data/nuscenes/samples/LIDAR_TOP/CAM_BACK_LEFT directory

* Add files via upload

* Remove unnecessary 'f'

* Remove unnecessary \ in arguments

* Remove check for pycocotools version because it has been done in the cocodataset

* Remove unnecessary comma, add TODO and change init of attrs in format_results

* Merge RandomFlip3D and RandomFlipMono3D

* Add pytest to check whether cuda is available in the unit test

* Add visualization TODO

* Remove useless init in loading mono3d images

- fix 'invalid configuration argument' error triggered by empty point input. test cases covering similar situations are added to test_dynamic_scatter.py as well.
trivial changes:
- switch to using torch::unique_dim to generate reduce mapping instead of calculating it from scratch.