Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[Feature] Adding PWC metafile #485

Merged
merged 7 commits into from
Apr 28, 2021
Merged

[Feature] Adding PWC metafile #485

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
e82326a
[Feature] Adding PWC metafile
WU-Wenhao Apr 26, 2021
dd06555
fix url in nuimages
WU-Wenhao Apr 26, 2021
a370148
removing inaccurate training time
WU-Wenhao Apr 27, 2021
dc09fb2
refine pwc index
WU-Wenhao Apr 27, 2021
ee9eca4
refine training resources
WU-Wenhao Apr 27, 2021
2bc2cef
removing empty metric of metafile in regnet
WU-Wenhao Apr 28, 2021
f97a82a
removing 'Focal_Loss' and 'SECOND' from Architecture
WU-Wenhao Apr 28, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
26 changes: 26 additions & 0 deletions configs/3dssd/metafile.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,26 @@
Collections:
- Name: 3DSSD
Metadata:
Training Data: KITTI
Training Techniques:
- AdamW
Training Resources: 8x V100 GPUs
Architecture:
- PointNet++
- SSD
ZwwWayne marked this conversation as resolved.
Show resolved Hide resolved
Paper: https://arxiv.org/abs/2002.10187
README: configs/3dssd/README.md

Models:
- Name: 3dssd_kitti-3d-car
In Collection: 3DSSD
Config: configs/3dssd/3dssd_kitti-3d-car.py
Metadata:
train time (s/iter): 0.42
Training Memory (GB): 4.7
Results:
- Task: 3D Object Detection
Dataset: KITTI
Metrics:
mAP: 78.39(81.00)
ZwwWayne marked this conversation as resolved.
Show resolved Hide resolved
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/3dssd/3dssd_kitti-3d-car_20210324_122002-07e9a19b.pth
97 changes: 97 additions & 0 deletions configs/centerpoint/metafile.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,97 @@
Collections:
- Name: CenterPoint
Metadata:
Training Data: nuScenes
Training Techniques:
- AdamW
Training Resources: 8x V100 GPUs
Architecture:
- Feature Pyramid Network
ZwwWayne marked this conversation as resolved.
Show resolved Hide resolved
- Hard Voxelization
Paper: https://arxiv.org/abs/2006.11275
README: configs/centerpoint/README.md

Models:
- Name: centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus
In Collection: CenterPoint
Config: configs/centerpoint/centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus.py
Metadata:
train time (s/iter): 3.02
Training Memory (GB): 4.9
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
mAP: 56.19
NDS: 64.43
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/centerpoint/centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus_20201001_135205-5db91e00.pth

- Name: centerpoint_01voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus
In Collection: CenterPoint
Config: configs/centerpoint/centerpoint_01voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus.py
Metadata:
train time (s/iter): 3.26
Training Memory (GB): 5.2
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
mAP: 56.34
NDS: 64.81
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/centerpoint/centerpoint_01voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus/centerpoint_01voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus_20201004_075317-26d8176c.pth

- Name: centerpoint_0075voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus
In Collection: CenterPoint
Config: configs/centerpoint/centerpoint_0075voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus.py
Metadata:
train time (s/iter): 3.06
Training Memory (GB): 7.8
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
mAP: 57.34
NDS: 65.23
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/centerpoint/centerpoint_0075voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_0075voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus_20200925_230905-358fbe3b.pth

- Name: centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus
In Collection: CenterPoint
Config: configs/centerpoint/centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus.py
Metadata:
train time (s/iter): 3.58
Training Memory (GB): 8.5
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
mAP: 57.27
NDS: 65.58
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/centerpoint/centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus/centerpoint_0075voxel_second_secfpn_dcn_circlenms_4x8_cyclic_20e_nus_20200930_201619-67c8496f.pth

- Name: centerpoint_02pillar_second_secfpn_circlenms_4x8_cyclic_20e_nus
In Collection: CenterPoint
Config: configs/centerpoint/centerpoint_02pillar_second_secfpn_circlenms_4x8_cyclic_20e_nus.py
Metadata:
train time (s/iter): 1.17
Training Memory (GB): 4.4
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
mAP: 49.07
NDS: 59.66
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/centerpoint/centerpoint_01voxel_second_secfpn_circlenms_4x8_cyclic_20e_nus/centerpoint_02pillar_second_secfpn_circlenms_4x8_cyclic_20e_nus_20201004_170716-a134a233.pth

- Name: centerpoint_02pillar_second_secfpn_dcn_4x8_cyclic_20e_nus
In Collection: CenterPoint
Config: configs/centerpoint/centerpoint_02pillar_second_secfpn_dcn_4x8_cyclic_20e_nus.py
Metadata:
train time (s/iter): 1.47
Training Memory (GB): 4.6
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
mAP: 48.8
NDS: 59.67
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/centerpoint/centerpoint_02pillar_second_secfpn_dcn_4x8_cyclic_20e_nus/centerpoint_02pillar_second_secfpn_dcn_4x8_cyclic_20e_nus_20200930_103722-3bb135f2.pth
53 changes: 53 additions & 0 deletions configs/dynamic_voxelization/metafile.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,53 @@
Collections:
- Name: Dynamic Voxelization
Metadata:
Training Data: KITTI
Training Techniques:
- AdamW
Training Resources: 8x V100 GPUs
Architecture:
- Feature Pyramid Network
ZwwWayne marked this conversation as resolved.
Show resolved Hide resolved
- Dynamic Voxelization
- Focal Loss
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is focal loss also used in other models?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Whether the loss should be regarded as a part of architecture?

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can omit Focal Loss and SECOND.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, sir. Please refer to metafile in fp16, free_anchor, imvotenet, mvxnet, parta2, pointpillars, regnet, second and ssn. Due to the high frequency, I thought focal loss is a common architecture and is written into metafile.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for your advice. I have updated the metafiles.

Paper: https://arxiv.org/abs/1910.06528
README: configs/dynamic_voxelization/README.md

Models:
- Name: dv_second_secfpn_6x8_80e_kitti-3d-car
In Collection: Dynamic Voxelization
Config: configs/dynamic_voxelization/dv_second_secfpn_6x8_80e_kitti-3d-car.py
Metadata:
train time (s/iter): 0.84
Training Memory (GB): 5.5
Results:
- Task: 3D Object Detection
Dataset: KITTI
Metrics:
mAP: 78.83
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/dynamic_voxelization/dv_second_secfpn_6x8_80e_kitti-3d-car/dv_second_secfpn_6x8_80e_kitti-3d-car_20200620_235228-ac2c1c0c.pth

- Name: dv_second_secfpn_2x8_cosine_80e_kitti-3d-3class
In Collection: Dynamic Voxelization
Config: configs/dynamic_voxelization/dv_second_secfpn_2x8_cosine_80e_kitti-3d-3class.py
Metadata:
train time (s/iter): 0.94
Training Memory (GB): 5.5
Results:
- Task: 3D Object Detection
Dataset: KITTI
Metrics:
mAP: 65.10
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/dynamic_voxelization/dv_second_secfpn_2x8_cosine_80e_kitti-3d-3class/dv_second_secfpn_2x8_cosine_80e_kitti-3d-3class_20200620_231010-6aa607d3.pth

- Name: dv_pointpillars_secfpn_6x8_160e_kitti-3d-car
In Collection: Dynamic Voxelization
Config: configs/dynamic_voxelization/dv_pointpillars_secfpn_6x8_160e_kitti-3d-car.py
Metadata:
train time (s/iter): 0.43
Training Memory (GB): 4.7
Results:
- Task: 3D Object Detection
Dataset: KITTI
Metrics:
mAP: 77.76
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/dynamic_voxelization/dv_pointpillars_secfpn_6x8_160e_kitti-3d-car/dv_pointpillars_secfpn_6x8_160e_kitti-3d-car_20200620_230844-ee7b75c9.pth
85 changes: 85 additions & 0 deletions configs/fp16/metafile.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,85 @@
Collections:
- Name: FP16
Metadata:
Training Data:
- KITTI
- nuScenes
Training Techniques:
- AdamW
ZwwWayne marked this conversation as resolved.
Show resolved Hide resolved
Training Resources: 8x TITAN Xp
Architecture:
- Hard Voxelization
- Focal Loss
Paper:
- https://www.mdpi.com/1424-8220/18/10/3337
- https://arxiv.org/abs/1812.05784
README: configs/fp16/README.md

Models:
- Name: hv_second_secfpn_fp16_6x8_80e_kitti-3d-car
In Collection: FP16
Config: configs/fp16/hv_second_secfpn_fp16_6x8_80e_kitti-3d-car.py
Metadata:
train time (s/iter): 0.84
Training Memory (GB):
- FP32: 5.4
- FP16: 2.9
Results:
- Task: 3D Object Detection
Dataset: KITTI
Metrics:
FP32 mAP: 79.07
FP16 mAP: 78.72
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/fp16/hv_second_secfpn_fp16_6x8_80e_kitti-3d-car/hv_second_secfpn_fp16_6x8_80e_kitti-3d-car_20200924_211301-1f5ad833.pth

- Name: hv_second_secfpn_fp16_6x8_80e_kitti-3d-3class
In Collection: FP16
Config: configs/fp16/hv_second_secfpn_fp16_6x8_80e_kitti-3d-3class.py
Metadata:
train time (s/iter): 0.98
Training Memory (GB):
- FP32: 5.4
- FP16: 2.9
Results:
- Task: 3D Object Detection
Dataset: KITTI
Metrics:
FP32 mAP: 64.41
FP16 mAP: 67.4
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/fp16/hv_second_secfpn_fp16_6x8_80e_kitti-3d-3class/hv_second_secfpn_fp16_6x8_80e_kitti-3d-3class_20200925_110059-05f67bdf.pth

- Name: hv_pointpillars_secfpn_sbn-all_fp16_2x8_2x_nus-3d
In Collection: FP16
Config: configs/fp16/hv_pointpillars_secfpn_sbn-all_fp16_2x8_2x_nus-3d.py
Metadata:
train time (s/iter): 2.14
Training Memory (GB):
- FP32: 16.4
- FP16: 8.37
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
FP32 mAP: 35.17
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

FP32 AP is just a reference of this model and is produced by its FP32 version. So we do not need to add it

FP32 NDS: 49.7
FP16 mAP: 35.19
FP16 NDS: 50.27
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/fp16/hv_pointpillars_secfpn_sbn-all_fp16_2x8_2x_nus-3d/hv_pointpillars_secfpn_sbn-all_fp16_2x8_2x_nus-3d_20201020_222626-c3f0483e.pth

- Name: hv_pointpillars_fpn_sbn-all_fp16_2x8_2x_nus-3d
In Collection: FP16
Config: configs/fp16/hv_pointpillars_fpn_sbn-all_fp16_2x8_2x_nus-3d.py
Metadata:
train time (s/iter): 2.21
Training Memory (GB):
- FP32: 16.4
- FP16: 8.40
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
FP32 mAP: 40.0
FP32 NDS: 53.3
FP16 mAP: 39.26
FP16 NDS: 53.26
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/fp16/hv_pointpillars_fpn_sbn-all_fp16_2x8_2x_nus-3d/hv_pointpillars_fpn_sbn-all_fp16_2x8_2x_nus-3d_20201021_120719-269f9dd6.pth
97 changes: 97 additions & 0 deletions configs/free_anchor/metafile.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,97 @@
Collections:
- Name: FreeAnchor
Metadata:
Training Data: nuScenes
Training Techniques:
- AdamW
Training Resources: 8x V100 GPUs
Architecture:
- Hard Voxelization
- Focal Loss
ZwwWayne marked this conversation as resolved.
Show resolved Hide resolved
Paper: https://arxiv.org/abs/1909.02466
README: configs/free_anchor/README.md

Models:
- Name: hv_pointpillars_fpn_sbn-all_free-anchor_4x8_2x_nus-3d
ZwwWayne marked this conversation as resolved.
Show resolved Hide resolved
In Collection: FreeAnchor
Config: free_anchor/hv_pointpillars_fpn_sbn-all_free-anchor_4x8_2x_nus-3d.py
Metadata:
train time (s/iter): 1.27
Training Memory (GB): 16.2
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
mAP: 43.7
NDS: 55.3
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/free_anchor/hv_pointpillars_fpn_sbn-all_free-anchor_4x8_2x_nus-3d/hv_pointpillars_fpn_sbn-all_free-anchor_4x8_2x_nus-3d_20200628_210537-09d359fc.pth

- Name: hv_pointpillars_regnet-400mf_fpn_sbn-all_free-anchor_4x8_2x_nus-3d
In Collection: FreeAnchor
Config: configs/free_anchor/hv_pointpillars_regnet-400mf_fpn_sbn-all_free-anchor_4x8_2x_nus-3d.py
Metadata:
train time (s/iter): 1.51
Training Memory (GB): 17.7
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
mAP: 47.9
NDS: 58.6
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/free_anchor/hv_pointpillars_regnet-400mf_fpn_sbn-all_free-anchor_4x8_2x_nus-3d/hv_pointpillars_regnet-400mf_fpn_sbn-all_free-anchor_4x8_2x_nus-3d_20200629_050311-a334765d.pth

- Name: hv_pointpillars_regnet-1.6gf_fpn_sbn-all_free-anchor_4x8_2x_nus-3d
In Collection: FreeAnchor
Config: configs/free_anchor/hv_pointpillars_regnet-1.6gf_fpn_sbn-all_free-anchor_4x8_2x_nus-3d.py
Metadata:
train time (s/iter): 1.82
Training Memory (GB): 24.3
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
mAP: 51.2
NDS: 60.8
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/free_anchor/hv_pointpillars_regnet-1.6gf_fpn_sbn-all_free-anchor_4x8_2x_nus-3d/hv_pointpillars_regnet-1.6gf_fpn_sbn-all_free-anchor_4x8_2x_nus-3d_20200629_105446-6ffa59cb.pth

- Name: hv_pointpillars_regnet-1.6gf_fpn_sbn-all_free-anchor_strong-aug_4x8_3x_nus-3d
In Collection: FreeAnchor
Config: configs/free_anchor/hv_pointpillars_regnet-1.6gf_fpn_sbn-all_free-anchor_strong-aug_4x8_3x_nus-3d.py
Metadata:
train time (s/iter): 1.82
Training Memory (GB): 24.3
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
mAP: 53.0
NDS: 62.2
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/free_anchor/hv_pointpillars_regnet-1.6gf_fpn_sbn-all_free-anchor_strong-aug_4x8_3x_nus-3d/hv_pointpillars_regnet-1.6gf_fpn_sbn-all_free-anchor_strong-aug_4x8_3x_nus-3d_20200701_201531-036f7de3.pth

- Name: hv_pointpillars_regnet-3.2gf_fpn_sbn-all_free-anchor_4x8_2x_nus-3d
In Collection: FreeAnchor
Config: configs/free_anchor/hv_pointpillars_regnet-3.2gf_fpn_sbn-all_free-anchor_4x8_2x_nus-3d.py
Metadata:
train time (s/iter): 2.16
Training Memory (GB): 29.5
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
mAP: 52.2
NDS: 62.0
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/free_anchor/hv_pointpillars_regnet-3.2gf_fpn_sbn-all_free-anchor_4x8_2x_nus-3d/hv_pointpillars_regnet-3.2gf_fpn_sbn-all_free-anchor_4x8_2x_nus-3d_20200629_055854-658125b0.pth

- Name: hv_pointpillars_regnet-3.2gf_fpn_sbn-all_free-anchor_strong-aug_4x8_3x_nus-3d
In Collection: FreeAnchor
Config: configs/free_anchor/hv_pointpillars_regnet-3.2gf_fpn_sbn-all_free-anchor_strong-aug_4x8_3x_nus-3d.py
Metadata:
train time (s/iter): 2.16
Training Memory (GB): 29.5
Results:
- Task: 3D Object Detection
Dataset: nuScenes
Metrics:
mAP: 55.09
NDS: 63.5
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/free_anchor/hv_pointpillars_regnet-3.2gf_fpn_sbn-all_free-anchor_strong-aug_4x8_3x_nus-3d/hv_pointpillars_regnet-3.2gf_fpn_sbn-all_free-anchor_strong-aug_4x8_3x_nus-3d_20200629_181452-297fdc66.pth
26 changes: 26 additions & 0 deletions configs/h3dnet/metafile.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,26 @@
Collections:
- Name: H3DNet
Metadata:
Training Data: ScanNet
Training Techniques:
- AdamW
Training Resources: 8x GeForce GTX 1080 Ti
Architecture:
- H3DRoI
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

can leave it empty

Paper: https://arxiv.org/abs/2006.05682
README: configs/h3dnet/README.md

Models:
- Name: h3dnet_3x8_scannet-3d-18class
In Collection: H3DNet
Config: configs/h3dnet/h3dnet_3x8_scannet-3d-18class.py
Metadata:
train time (s/iter): 2.73
Training Memory (GB): 7.9
Results:
- Task: 3D Object Detection
Dataset: ScanNet
Metrics:
AP@0.25: 66.43
AP@0.5: 48.01
Weights: https://download.openmmlab.com/mmdetection3d/v0.1.0_models/h3dnet/h3dnet_scannet-3d-18class/h3dnet_scannet-3d-18class_20200830_000136-02e36246.pth
Loading