Skip to content

Commit f4b0174

Browse files
xizaoquxizaoqu
authored
[Docs] Add docs and README for Cylinder3D (#2350)
* add readme * fix * fix * fix * rebase * add link * add link * add link
1 parent 3e2de53 commit f4b0174

File tree

6 files changed

+128
-51
lines changed

6 files changed

+128
-51
lines changed

‎README.md

Lines changed: 31 additions & 25 deletions
Original file line numberDiff line numberDiff line change
@@ -134,6 +134,7 @@ Results and models are available in the [model zoo](docs/en/model_zoo.md).
134134
<li><a href="configs/dgcnn">DGCNN (TOG'2019)</a></li>
135135
<li>DLA (CVPR'2018)</li>
136136
<li>MinkResNet (CVPR'2019)</li>
137+
<li><a href="configs/cylinder3d">Cylinder3D (CVPR'2021)</a></li>
137138
</ul>
138139
</td>
139140
<td>
@@ -218,6 +219,10 @@ Results and models are available in the [model zoo](docs/en/model_zoo.md).
218219
</ul>
219220
</td>
220221
<td>
222+
<li><b>Outdoor</b></li>
223+
<ul>
224+
<li><a href="configs/cylinder3d">Cylinder3D (CVPR'2021)</a></li>
225+
</ul>
221226
<li><b>Indoor</b></li>
222227
<ul>
223228
<li><a href="configs/pointnet2">PointNet++ (NeurIPS'2017)</a></li>
@@ -232,31 +237,32 @@ Results and models are available in the [model zoo](docs/en/model_zoo.md).
232237
</tbody>
233238
</table>
234239

235-
| | ResNet | PointNet++ | SECOND | DGCNN | RegNetX | DLA | MinkResNet |
236-
| :-----------: | :----: | :--------: | :----: | :---: | :-----: | :-: | :--------: |
237-
| SECOND | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ |
238-
| PointPillars | ✗ | ✗ | ✓ | ✗ | ✓ | ✗ | ✗ |
239-
| FreeAnchor | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ | ✗ |
240-
| VoteNet | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
241-
| H3DNet | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
242-
| 3DSSD | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
243-
| Part-A2 | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ |
244-
| MVXNet | ✓ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ |
245-
| CenterPoint | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ |
246-
| SSN | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ | ✗ |
247-
| ImVoteNet | ✓ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
248-
| FCOS3D | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
249-
| PointNet++ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
250-
| Group-Free-3D | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
251-
| ImVoxelNet | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
252-
| PAConv | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
253-
| DGCNN | ✗ | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ |
254-
| SMOKE | ✗ | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ |
255-
| PGD | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
256-
| MonoFlex | ✗ | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ |
257-
| SA-SSD | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ |
258-
| FCAF3D | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✓ |
259-
| PV-RCNN | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ |
240+
| | ResNet | PointNet++ | SECOND | DGCNN | RegNetX | DLA | MinkResNet | Cylinder3D |
241+
| :-----------: | :----: | :--------: | :----: | :---: | :-----: | :-: | :--------: | :--------: |
242+
| SECOND | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
243+
| PointPillars | ✗ | ✗ | ✓ | ✗ | ✓ | ✗ | ✗ | ✗ |
244+
| FreeAnchor | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ |
245+
| VoteNet | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
246+
| H3DNet | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
247+
| 3DSSD | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
248+
| Part-A2 | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
249+
| MVXNet | ✓ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
250+
| CenterPoint | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
251+
| SSN | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ |
252+
| ImVoteNet | ✓ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
253+
| FCOS3D | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
254+
| PointNet++ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
255+
| Group-Free-3D | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
256+
| ImVoxelNet | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
257+
| PAConv | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
258+
| DGCNN | ✗ | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ |
259+
| SMOKE | ✗ | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ | ✗ |
260+
| PGD | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
261+
| MonoFlex | ✗ | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ | ✗ |
262+
| SA-SSD | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
263+
| FCAF3D | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ |
264+
| PV-RCNN | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
265+
| Cylinder3D | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✓ |
260266

261267
**Note:** All the about **300+ models, methods of 40+ papers** in 2D detection supported by [MMDetection](https://github.com/open-mmlab/mmdetection/blob/3.x/docs/en/model_zoo.md) can be trained or used in this codebase.
262268

‎README_zh-CN.md

Lines changed: 31 additions & 25 deletions
Original file line numberDiff line numberDiff line change
@@ -131,6 +131,7 @@ MMDetection3D 是一个基于 PyTorch 的目标检测开源工具箱，下一代
131131
<li><a href="configs/dgcnn">DGCNN (TOG'2019)</a></li>
132132
<li>DLA (CVPR'2018)</li>
133133
<li>MinkResNet (CVPR'2019)</li>
134+
<li><a href="configs/cylinder3d">Cylinder3D (CVPR'2021)</a></li>
134135
</ul>
135136
</td>
136137
<td>
@@ -214,6 +215,10 @@ MMDetection3D 是一个基于 PyTorch 的目标检测开源工具箱，下一代
214215
</ul>
215216
</td>
216217
<td>
218+
<li><b>室外</b></li>
219+
<ul>
220+
<li><a href="configs/cylinder3d">Cylinder3D (CVPR'2021)</a></li>
221+
</ul>
217222
<li><b>室内</b></li>
218223
<ul>
219224
<li><a href="configs/pointnet2">PointNet++ (NeurIPS'2017)</a></li>
@@ -228,31 +233,32 @@ MMDetection3D 是一个基于 PyTorch 的目标检测开源工具箱，下一代
228233
</tbody>
229234
</table>
230235

231-
| | ResNet | PointNet++ | SECOND | DGCNN | RegNetX | DLA | MinkResNet |
232-
| :-----------: | :----: | :--------: | :----: | :---: | :-----: | :-: | :--------: |
233-
| SECOND | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ |
234-
| PointPillars | ✗ | ✗ | ✓ | ✗ | ✓ | ✗ | ✗ |
235-
| FreeAnchor | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ | ✗ |
236-
| VoteNet | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
237-
| H3DNet | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
238-
| 3DSSD | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
239-
| Part-A2 | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ |
240-
| MVXNet | ✓ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ |
241-
| CenterPoint | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ |
242-
| SSN | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ | ✗ |
243-
| ImVoteNet | ✓ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
244-
| FCOS3D | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
245-
| PointNet++ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
246-
| Group-Free-3D | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
247-
| ImVoxelNet | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
248-
| PAConv | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
249-
| DGCNN | ✗ | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ |
250-
| SMOKE | ✗ | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ |
251-
| PGD | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
252-
| MonoFlex | ✗ | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ |
253-
| SA-SSD | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ |
254-
| FCAF3D | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✓ |
255-
| PV-RCNN | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ |
236+
| | ResNet | PointNet++ | SECOND | DGCNN | RegNetX | DLA | MinkResNet | Cylinder3D |
237+
| :-----------: | :----: | :--------: | :----: | :---: | :-----: | :-: | :--------: | :--------: |
238+
| SECOND | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
239+
| PointPillars | ✗ | ✗ | ✓ | ✗ | ✓ | ✗ | ✗ | ✗ |
240+
| FreeAnchor | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ |
241+
| VoteNet | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
242+
| H3DNet | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
243+
| 3DSSD | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
244+
| Part-A2 | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
245+
| MVXNet | ✓ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
246+
| CenterPoint | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
247+
| SSN | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ |
248+
| ImVoteNet | ✓ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
249+
| FCOS3D | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
250+
| PointNet++ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
251+
| Group-Free-3D | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
252+
| ImVoxelNet | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
253+
| PAConv | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
254+
| DGCNN | ✗ | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ |
255+
| SMOKE | ✗ | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ | ✗ |
256+
| PGD | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ |
257+
| MonoFlex | ✗ | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ | ✗ |
258+
| SA-SSD | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
259+
| FCAF3D | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✓ | ✗ |
260+
| PV-RCNN | ✗ | ✗ | ✓ | ✗ | ✗ | ✗ | ✗ | ✗ |
261+
| Cylinder3D | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✗ | ✓ |
256262

257263
**注意：**[MMDetection](https://github.com/open-mmlab/mmdetection/blob/3.x/docs/zh_cn/model_zoo.md) 支持的基于 2D 检测的 **300+ 个模型，40+ 的论文算法**在 MMDetection3D 中都可以被训练或使用。
258264

‎configs/cylinder3d/README.md

Lines changed: 35 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,35 @@
1+
# Cylindrical and Asymmetrical 3D Convolution Networks for LiDAR Segmentation
2+
3+
> [Cylindrical and Asymmetrical 3D Convolution Networks for LiDAR Segmentation](https://arxiv.org/abs/2011.10033)
4+
5+
<!-- [ALGORITHM] -->
6+
7+
## Abstract
8+
9+
State-of-the-art methods for large-scale driving-scene LiDAR segmentation often project the point clouds to 2D space and then process them via 2D convolution. Although this corporation shows the competitiveness in the point cloud, it inevitably alters and abandons the 3D topology and geometric relations. A natural remedy is to utilize the3D voxelization and 3D convolution network. However, we found that in the outdoor point cloud, the improvement obtained in this way is quite limited. An important reason is the property of the outdoor point cloud, namely sparsity and varying density. Motivated by this investigation, we propose a new framework for the outdoor LiDAR segmentation, where cylindrical partition and asymmetrical 3D convolution networks are designed to explore the 3D geometric pat-tern while maintaining these inherent properties. Moreover, a point-wise refinement module is introduced to alleviate the interference of lossy voxel-based label encoding. We evaluate the proposed model on two large-scale datasets, i.e., SemanticKITTI and nuScenes. Our method achieves the 1st place in the leaderboard of SemanticKITTI and outperforms existing methods on nuScenes with a noticeable margin, about 4%. Furthermore, the proposed 3D framework also generalizes well to LiDAR panoptic segmentation and LiDAR 3D detection.
10+
11+
## Introduction
12+
13+
We implement Cylinder3D and provide the result and checkpoints on Semantickitti datasets.
14+
15+
## Results and models
16+
17+
### SemanticKITTI
18+
19+
| Method | Lr schd | Mem (GB) | mIOU | Download |
20+
| :--------: | :-----: | :------: | :------: | :--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
21+
| Cylinder3D | 3x | 10.2 | 63.1±0.5 | [model](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/cylinder3d/cylinder3d_4xb4_3x_semantickitti/cylinder3d_4xb4_3x_semantickitti_20230318_191107.json) \| [log](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/cylinder3d/cylinder3d_4xb4_3x_semantickitti/cylinder3d_4xb4_3x_semantickitti_20230318_191107-822a8c31.pth) |
22+
23+
Note: We reproduce the performance comparable with its official repo (https://github.com/xinge008/Cylinder3D). It's slightly lower than the performance (65.9 mIOU) reported in the paper due to the lack of point-wise refinement and shorter training time.
24+
25+
## Citation
26+
27+
```latex
28+
@inproceedings{zhu2021cylindrical,
29+
title={Cylindrical and asymmetrical 3d convolution networks for lidar segmentation},
30+
author={Zhu, Xinge and Zhou, Hui and Wang, Tai and Hong, Fangzhou and Ma, Yuexin and Li, Wei and Li, Hongsheng and Lin, Dahua},
31+
booktitle={Proceedings of the IEEE/CVF conference on computer vision and pattern recognition},
32+
pages={9939--9948},
33+
year={2021}
34+
}
35+
```

‎configs/cylinder3d/metafile.yml

Lines changed: 29 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,29 @@
1+
Collections:
2+
- Name: Cylinder3D
3+
Metadata:
4+
Training Techniques:
5+
- AdamW
6+
Training Resources: 4x A100 GPUs
7+
Architecture:
8+
- Cylinder3D
9+
Paper:
10+
URL: https://arxiv.org/abs/2011.10033
11+
Title: 'Cylindrical and Asymmetrical 3D Convolution Networks for LiDAR Segmentation'
12+
README: configs/cylinder3d/README.md
13+
Code:
14+
URL: https://github.com/open-mmlab/mmdetection3d/blob/dev-1.x/mmdet3d/models/segmentors/cylinder3d.py#L13
15+
Version: v1.1.0rc4
16+
17+
Models:
18+
- Name:
19+
In Collection: Cylinder3D
20+
Config: configs/cylinder3d/cylinder3d_4xb4_3x_semantickitti.py
21+
Metadata:
22+
Training Data: SemanticKITTI
23+
Training Memory (GB): 10.2
24+
Results:
25+
- Task: 3D Semantic Segmentation
26+
Dataset: SemanticKITTI
27+
Metrics:
28+
mIOU: 63.1
29+
Weights:

‎mmdet3d/datasets/semantickitti_dataset.py

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -21,7 +21,7 @@ class SemanticKittiDataset(Seg3DDataset):
2121
metainfo (dict, optional): Meta information for dataset, such as class
2222
information. Defaults to None.
2323
data_prefix (dict): Prefix for training data. Defaults to
24-
dict(pts='points',
24+
dict(pts='',
2525
img='',
2626
pts_instance_mask='',
2727
pts_semantic_mask='').

‎model-index.yml

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -23,5 +23,6 @@ Import:
2323
- configs/smoke/metafile.yml
2424
- configs/ssn/metafile.yml
2525
- configs/votenet/metafile.yml
26+
- configs/cylinder3d/metafile.yml
2627
- configs/pv_rcnn/metafile.yml
2728
- configs/fcaf3d/metafile.yml

0 commit comments

Comments
 (0)