Skip to content

Commit 56657c2

Browse files
sunjiahao1999sunjiahao1999
authored
[Docs] Update readme of lidar segmentation methods (#2559)
* add readme * fix spvcnn memory * add fps and training time * resolve typo * fix typo
1 parent fa724b1 commit 56657c2

File tree

6 files changed

+152
-29
lines changed

6 files changed

+152
-29
lines changed

‎configs/cylinder3d/README.md

Lines changed: 4 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -18,9 +18,10 @@ We implement Cylinder3D and provide the result and checkpoints on Semantickitti
1818

1919
### SemanticKITTI
2020

21-
| Method | Lr schd | Mem (GB) | mIOU | Download |
22-
| :--------: | :-----: | :------: | :------: | :--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
23-
| 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_191107-822a8c31.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/cylinder3d/cylinder3d_4xb4_3x_semantickitti/cylinder3d_4xb4_3x_semantickitti_20230318_191107.json) |
21+
| Method | Lr schd | Laser-Polar Mix | Mem (GB) | mIoU | Download |
22+
| :-----------------------------------------------------------------: | :-----: | :-------------: | :------: | :------: | :--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
23+
| [Cylinder3D](./cylinder3d_8xb2-laser-polar-mix-3x_semantickitti.py) | 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_191107-822a8c31.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/cylinder3d/cylinder3d_4xb4_3x_semantickitti/cylinder3d_4xb4_3x_semantickitti_20230318_191107.json) |
24+
| [Cylinder3D](./cylinder3d_8xb2-laser-polar-mix-3x_semantickitti.py) | 3x | ✔ | 12.8 | 67.0 | [model](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/cylinder3d/cylinder3d_8xb2-amp-laser-polar-mix-3x_semantickitti_20230425_144950-372cdf69.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/cylinder3d/cylinder3d_8xb2-amp-laser-polar-mix-3x_semantickitti_20230425_144950.log) |
2425

2526
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.
2627

‎configs/cylinder3d/metafile.yml

Lines changed: 16 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -15,7 +15,7 @@ Collections:
1515
Version: v1.1.0
1616

1717
Models:
18-
- Name:
18+
- Name: cylinder3d_4xb4-3x_semantickitti
1919
In Collection: Cylinder3D
2020
Config: configs/cylinder3d/cylinder3d_4xb4_3x_semantickitti.py
2121
Metadata:
@@ -25,5 +25,18 @@ Models:
2525
- Task: 3D Semantic Segmentation
2626
Dataset: SemanticKITTI
2727
Metrics:
28-
mIOU: 63.1
29-
Weights:
28+
mIoU: 63.1
29+
Weights: https://download.openmmlab.com/mmdetection3d/v1.1.0_models/cylinder3d/cylinder3d_4xb4_3x_semantickitti/cylinder3d_4xb4_3x_semantickitti_20230318_191107-822a8c31.pth
30+
31+
- Name: cylinder3d_8xb2-laser-polar-mix-3x_semantickitti
32+
In Collection: Cylinder3D
33+
Config: configs/cylinder3d/cylinder3d_8xb2-laser-polar-mix-3x_semantickitti.py
34+
Metadata:
35+
Training Data: SemanticKITTI
36+
Training Memory (GB): 12.8
37+
Results:
38+
- Task: 3D Semantic Segmentation
39+
Dataset: SemanticKITTI
40+
Metrics:
41+
mIoU: 67.0
42+
Weights: https://download.openmmlab.com/mmdetection3d/v1.1.0_models/cylinder3d/cylinder3d_4xb4_3x_semantickitti/cylinder3d_4xb4_3x_semantickitti_20230318_191107-822a8c31.pth

‎configs/minkunet/README.md

Lines changed: 16 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -14,22 +14,32 @@ In many robotics and VR/AR applications, 3D-videos are readily-available sources
1414

1515
## Introduction
1616

17-
We implement MinkUNet with [TorchSparse](https://github.com/mit-han-lab/torchsparse) backend and provide the result and checkpoints on SemanticKITTI datasets.
17+
We implement MinkUNet with [TorchSparse](https://github.com/mit-han-lab/torchsparse) / [Minkowski Engine](https://github.com/NVIDIA/MinkowskiEngine) / [Spconv](https://github.com/traveller59/spconv) backend and provide the result and checkpoints on SemanticKITTI datasets.
1818

1919
## Results and models
2020

2121
### SemanticKITTI
2222

23-
| Method | Lr schd | Mem (GB) | mIoU | Download |
24-
| :----------: | :-----: | :------: | :--: | :---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
25-
| MinkUNet-W16 | 15e | 3.4 | 60.3 | [model](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet_w16_8xb2-15e_semantickitti/minkunet_w16_8xb2-15e_semantickitti_20230309_160737-0d8ec25b.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet_w16_8xb2-15e_semantickitti/minkunet_w16_8xb2-15e_semantickitti_20230309_160737.log) |
26-
| MinkUNet-W20 | 15e | 3.7 | 61.6 | [model](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet_w20_8xb2-15e_semantickitti/minkunet_w20_8xb2-15e_semantickitti_20230309_160718-c3b92e6e.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet_w20_8xb2-15e_semantickitti/minkunet_w20_8xb2-15e_semantickitti_20230309_160718.log) |
27-
| MinkUNet-W32 | 15e | 4.9 | 63.1 | [model](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet_w32_8xb2-15e_semantickitti/minkunet_w32_8xb2-15e_semantickitti_20230309_160710-7fa0a6f1.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet_w32_8xb2-15e_semantickitti/minkunet_w32_8xb2-15e_semantickitti_20230309_160710.log) |
23+
| Method | Backend | Lr schd | Amp | Laser-Polar Mix | Mem (GB) | Training Time (hours) | FPS | mIoU | Download |
24+
| :-------------------------------------------------------------------------------------------: | :--------------: | :-----: | :-: | :-------------: | :------: | :-------------------: | :----: | :--: | :---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------: |
25+
| [MinkUNet18-W16](./minkunet18_w16_torchsparse_8xb2-amp-15e_semantickitti.py) | torchsparse | 15e | ✔ | ✗ | 3.4 | - | - | 60.3 | [model](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet_w16_8xb2-15e_semantickitti/minkunet_w16_8xb2-15e_semantickitti_20230309_160737-0d8ec25b.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet_w16_8xb2-15e_semantickitti/minkunet_w16_8xb2-15e_semantickitti_20230309_160737.log) |
26+
| [MinkUNet18-W20](./minkunet18_w20_torchsparse_8xb2-amp-15e_semantickitti.py) | torchsparse | 15e | ✔ | ✗ | 3.7 | - | - | 61.6 | [model](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet_w20_8xb2-15e_semantickitti/minkunet_w20_8xb2-15e_semantickitti_20230309_160718-c3b92e6e.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet_w20_8xb2-15e_semantickitti/minkunet_w20_8xb2-15e_semantickitti_20230309_160718.log) |
27+
| [MinkUNet18-W32](./minkunet18_w32_torchsparse_8xb2-amp-15e_semantickitti.py) | torchsparse | 15e | ✔ | ✗ | 4.9 | - | - | 63.1 | [model](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet_w32_8xb2-15e_semantickitti/minkunet_w32_8xb2-15e_semantickitti_20230309_160710-7fa0a6f1.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet_w32_8xb2-15e_semantickitti/minkunet_w32_8xb2-15e_semantickitti_20230309_160710.log) |
28+
| [MinkUNet34-W32](./minkunet34_w32_minkowski_8xb2-laser-polar-mix-3x_semantickitti.py) | minkowski engine | 3x | ✗ | ✔ | 11.5 | 6.5 | 12.2 | 69.2 | [model](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet34_w32_minkowski_8xb2-laser-polar-mix-3x_semantickitti_20230514_202236-839847a8.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet34_w32_minkowski_8xb2-laser-polar-mix-3x_semantickitti_20230514_202236.log) |
29+
| [MinkUNet34-W32](./minkunet34_w32_spconv_8xb2-amp-laser-polar-mix-3x_semantickitti.py) | spconv | 3x | ✔ | ✔ | 6.7 | 2 | 14.6\* | 68.3 | [model](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet34_w32_spconv_8xb2-amp-laser-polar-mix-3x_semantickitti_20230512_233152-e0698a0f.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet34_w32_spconv_8xb2-amp-laser-polar-mix-3x_semantickitti_20230512_233152.log) |
30+
| [MinkUNet34-W32](./minkunet34_w32_spconv_8xb2-laser-polar-mix-3x_semantickitti.py) | spconv | 3x | ✗ | ✔ | 10.5 | 6 | 14.5 | 3 | 69.3 |
31+
| [MinkUNet34-W32](./minkunet34_w32_torchsparse_8xb2-amp-laser-polar-mix-3x_semantickitti.py) | torchsparse | 3x | ✔ | ✔ | 6.6 | 3 | 12.8 | 69.3 | [model](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet34_w32_torchsparse_8xb2-amp-laser-polar-mix-3x_semantickitti_20230512_233511-bef6cad0.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet34_w32_torchsparse_8xb2-amp-laser-polar-mix-3x_semantickitti_20230512_233511.log) |
32+
| [MinkUNet34-W32](./minkunet34_w32_torchsparse_8xb2-laser-polar-mix-3x_semantickitti.py) | torchsparse | 3x | ✗ | ✔ | 11.8 | 5.5 | 15.9 | 68.7 | [model](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet34_w32_torchsparse_8xb2-laser-polar-mix-3x_semantickitti_20230512_233601-2b61b0ab.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet34_w32_torchsparse_8xb2-laser-polar-mix-3x_semantickitti_20230512_233601.log) |
33+
| [MinkUNet34v2-W32](minkunet34v2_w32_torchsparse_8xb2-amp-laser-polar-mix-3x_semantickitti.py) | torchsparse | 3x | ✔ | ✔ | 8.9 | - | - | 70.3 | [model](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet34v2_w32_torchsparse_8xb2-amp-laser-polar-mix-3x_semantickitti_20230510_221853-b14a68b3.pth) \| [log](https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet34v2_w32_torchsparse_8xb2-amp-laser-polar-mix-3x_semantickitti_20230510_221853.log) |
2834

2935
**Note:** We follow the implementation in SPVNAS original [repo](https://github.com/mit-han-lab/spvnas) and W16\\W20\\W32 indicates different number of channels.
3036

3137
**Note:** Due to TorchSparse backend, the model performance is unstable with TorchSparse backend and may fluctuate by about 1.5 mIoU for different random seeds.
3238

39+
**Note:** Referring to [PCSeg](https://github.com/PJLab-ADG/PCSeg), MinkUNet34v2 is modified based on MinkUNet34.
40+
41+
**Note\*:** Training Time and FPS are measured on NVIDIA A100. The versions of Torchsparse, Minkowski Engine and Spconv are 0.5.4, 1.4.0 and 2.3.6 respectively. Since spconv 2.3.6 has a bug with fp16 on in the inference stage, the actual FPS measurement using fp32.
42+
3343
## Citation
3444

3545
```latex

‎configs/minkunet/metafile.yml

Lines changed: 90 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -14,9 +14,9 @@ Collections:
1414
Version: v1.1.0
1515

1616
Models:
17-
- Name: minkunet_w16_8xb2-15e_semantickitti
17+
- Name: minkunet18_w16_torchsparse_8xb2-amp-15e_semantickitti
1818
In Collection: MinkUNet
19-
Config: configs/minkunet/minkunet_w16_8xb2-15e_semantickitti.py
19+
Config: configs/minkunet/minkunet18_w16_torchsparse_8xb2-amp-15e_semantickitti.py
2020
Metadata:
2121
Training Data: SemanticKITTI
2222
Training Memory (GB): 3.4
@@ -28,9 +28,9 @@ Models:
2828
mIoU: 60.3
2929
Weights: https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet_w16_8xb2-15e_semantickitti/minkunet_w16_8xb2-15e_semantickitti_20230309_160737-0d8ec25b.pth
3030

31-
- Name: minkunet_w20_8xb2-15e_semantickitti
31+
- Name: minkunet18_w20_torchsparse_8xb2-amp-15e_semantickitti
3232
In Collection: MinkUNet
33-
Config: configs/minkunet/minkunet_w20_8xb2-15e_semantickitti.py
33+
Config: configs/minkunet/minkunet18_w20_torchsparse_8xb2-amp-15e_semantickitti.py
3434
Metadata:
3535
Training Data: SemanticKITTI
3636
Training Memory (GB): 3.7
@@ -42,9 +42,9 @@ Models:
4242
mIoU: 61.6
4343
Weights: https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet_w20_8xb2-15e_semantickitti/minkunet_w20_8xb2-15e_semantickitti_20230309_160718-c3b92e6e.pth
4444

45-
- Name: minkunet_w32_8xb2-15e_semantickitti
45+
- Name: minkunet18_w32_torchsparse_8xb2-amp-15e_semantickitti
4646
In Collection: MinkUNet
47-
Config: configs/minkunet/minkunet_w32_8xb2-15e_semantickitti.py
47+
Config: configs/minkunet/minkunet18_w32_torchsparse_8xb2-amp-15e_semantickitti.py
4848
Metadata:
4949
Training Data: SemanticKITTI
5050
Training Memory (GB): 4.9
@@ -55,3 +55,87 @@ Models:
5555
Metrics:
5656
mIoU: 63.1
5757
Weights: https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet_w32_8xb2-15e_semantickitti/minkunet_w32_8xb2-15e_semantickitti_20230309_160710-7fa0a6f1.pth
58+
59+
- Name: minkunet34_w32_minkowski_8xb2-laser-polar-mix-3x_semantickitti
60+
In Collection: MinkUNet
61+
Config: configs/minkunet/minkunet34_w32_minkowski_8xb2-laser-polar-mix-3x_semantickitti.py
62+
Metadata:
63+
Training Data: SemanticKITTI
64+
Training Memory (GB): 11.5
65+
Training Resources: 8x A100 GPUs
66+
Results:
67+
- Task: 3D Semantic Segmentation
68+
Dataset: SemanticKITTI
69+
Metrics:
70+
mIoU: 69.2
71+
Weights: https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet34_w32_minkowski_8xb2-laser-polar-mix-3x_semantickitti_20230514_202236-839847a8.pth
72+
73+
- Name: minkunet34_w32_spconv_8xb2-amp-laser-polar-mix-3x_semantickitti
74+
In Collection: MinkUNet
75+
Config: configs/minkunet/minkunet34_w32_spconv_8xb2-amp-laser-polar-mix-3x_semantickitti.py
76+
Metadata:
77+
Training Data: SemanticKITTI
78+
Training Memory (GB): 6.7
79+
Training Resources: 8x A100 GPUs
80+
Results:
81+
- Task: 3D Semantic Segmentation
82+
Dataset: SemanticKITTI
83+
Metrics:
84+
mIoU: 68.3
85+
Weights: https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet34_w32_spconv_8xb2-amp-laser-polar-mix-3x_semantickitti_20230512_233152-e0698a0f.pth
86+
87+
- Name: minkunet34_w32_spconv_8xb2-laser-polar-mix-3x_semantickitti
88+
In Collection: MinkUNet
89+
Config: configs/minkunet/minkunet34_w32_spconv_8xb2-laser-polar-mix-3x_semantickitti.py
90+
Metadata:
91+
Training Data: SemanticKITTI
92+
Training Memory (GB): 10.5
93+
Training Resources: 8x A100 GPUs
94+
Results:
95+
- Task: 3D Semantic Segmentation
96+
Dataset: SemanticKITTI
97+
Metrics:
98+
mIoU: 69.3
99+
Weights: https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet34_w32_spconv_8xb2-laser-polar-mix-3x_semantickitti_20230512_233817-72b200d8.pth
100+
101+
- Name: minkunet34_w32_torchsparse_8xb2-amp-laser-polar-mix-3x_semantickitti
102+
In Collection: MinkUNet
103+
Config: configs/minkunet/minkunet34_w32_torchsparse_8xb2-amp-laser-polar-mix-3x_semantickitti.py
104+
Metadata:
105+
Training Data: SemanticKITTI
106+
Training Memory (GB): 6.6
107+
Training Resources: 8x A100 GPUs
108+
Results:
109+
- Task: 3D Semantic Segmentation
110+
Dataset: SemanticKITTI
111+
Metrics:
112+
mIoU: 69.3
113+
Weights: https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet34_w32_torchsparse_8xb2-amp-laser-polar-mix-3x_semantickitti_20230512_233511-bef6cad0.pth
114+
115+
- Name: minkunet34_w32_torchsparse_8xb2-laser-polar-mix-3x_semantickitti
116+
In Collection: MinkUNet
117+
Config: configs/minkunet/minkunet34_w32_torchsparse_8xb2-laser-polar-mix-3x_semantickitti.py
118+
Metadata:
119+
Training Data: SemanticKITTI
120+
Training Memory (GB): 11.8
121+
Training Resources: 8x A100 GPUs
122+
Results:
123+
- Task: 3D Semantic Segmentation
124+
Dataset: SemanticKITTI
125+
Metrics:
126+
mIoU: 68.7
127+
Weights: https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet34_w32_torchsparse_8xb2-laser-polar-mix-3x_semantickitti_20230512_233601-2b61b0ab.pth
128+
129+
- Name: minkunet34v2_w32_torchsparse_8xb2-amp-laser-polar-mix-3x_semantickitti
130+
In Collection: MinkUNet
131+
Config: configs/minkunet/minkunet34v2_w32_torchsparse_8xb2-amp-laser-polar-mix-3x_semantickitti.py
132+
Metadata:
133+
Training Data: SemanticKITTI
134+
Training Memory (GB): 8.9
135+
Training Resources: 8x A100 GPUs
136+
Results:
137+
- Task: 3D Semantic Segmentation
138+
Dataset: SemanticKITTI
139+
Metrics:
140+
mIoU: 70.3
141+
Weights: https://download.openmmlab.com/mmdetection3d/v1.1.0_models/minkunet/minkunet34v2_w32_torchsparse_8xb2-amp-laser-polar-mix-3x_semantickitti_20230510_221853-b14a68b3.pth

0 commit comments

Comments
 (0)