Skip to content

Commit

Permalink
release pre-trained model for semantic segmentation
Browse files Browse the repository at this point in the history
  • Loading branch information
@yanx27
yanx27 committed Mar 26, 2021
1 parent 245c769 commit c28e1b6
Show file tree
Hide file tree
Showing 12 changed files with 5,532 additions and 1 deletion.
11 changes: 11 additions & 0 deletions README.md
View file
Original file line number Diff line number Diff line change
Expand Up @@ -3,6 +3,10 @@
This repo is implementation for [PointNet](http://openaccess.thecvf.com/content_cvpr_2017/papers/Qi_PointNet_Deep_Learning_CVPR_2017_paper.pdf) and [PointNet++](http://papers.nips.cc/paper/7095-pointnet-deep-hierarchical-feature-learning-on-point-sets-in-a-metric-space.pdf) in pytorch.

## Update
**2021/03/26:**

Release pre-trained models for semantic segmentation, where PointNet++ can achieve 53.5% for mIoU.

**2021/03/20:** Update codes for classification, including:

(1) Add codes for training **ModelNet10** dataset. Using setting of ``--num_category 10``.
Expand Down Expand Up @@ -107,6 +111,13 @@ python test_semseg.py --log_dir pointnet2_sem_seg --test_area 5 --visual
```
Visualization results will save in `log/sem_seg/pointnet2_sem_seg/visual/` and you can visualize these .obj file by [MeshLab](http://www.meshlab.net/).

### Performance
|Model | Overall Acc |Class avg IoU | Checkpoint
|--|--|--|--|
| PointNet (Official) | N/A | 41.1| - |
| PointNet (Pytorch) | 78.9 | 43.7| [40.7MB](log/sem_seg/pointnet_sem_seg) |
| PointNet2 (Official) | N/A |N/A | - |
| PointNet2_ssg (Pytorch) | 83.0 | **53.5**| [11.2MB](log/sem_seg/pointnet2_sem_seg) |

## Visualization
### Using show3d_balls.py
Expand Down
Binary file added log/sem_seg/pointnet2_sem_seg/checkpoints/best_model.pth
View file
Binary file not shown.
90 changes: 90 additions & 0 deletions log/sem_seg/pointnet2_sem_seg/eval.txt
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,90 @@
2021-03-22 13:40:15,976 - Model - INFO - PARAMETER ...
2021-03-22 13:40:15,977 - Model - INFO - Namespace(batch_size=32, gpu='2', log_dir='pointnet2_sem_seg', num_point=4096, num_votes=3, test_area=5, visual=False)
2021-03-22 13:45:50,923 - Model - INFO - The number of test data is: 68
2021-03-22 13:46:06,501 - Model - INFO - ---- EVALUATION WHOLE SCENE----
2021-03-22 13:54:10,224 - Model - INFO - Mean IoU of Area_5_office_1: 0.5885
2021-03-22 14:01:33,975 - Model - INFO - Mean IoU of Area_5_office_10: 0.6179
2021-03-22 14:38:11,357 - Model - INFO - Mean IoU of Area_5_conferenceRoom_3: 0.6024
2021-03-22 14:59:34,010 - Model - INFO - Mean IoU of Area_5_office_15: 0.5891
2021-03-22 15:24:52,909 - Model - INFO - Mean IoU of Area_5_lobby_1: 0.7761
2021-03-22 15:30:55,276 - Model - INFO - Mean IoU of Area_5_office_27: 0.6575
2021-03-22 16:07:05,579 - Model - INFO - Mean IoU of Area_5_hallway_13: 0.6365
2021-03-22 16:38:57,175 - Model - INFO - Mean IoU of Area_5_office_41: 0.6261
2021-03-22 16:57:27,187 - Model - INFO - Mean IoU of Area_5_office_18: 0.6153
2021-03-22 17:06:25,938 - Model - INFO - Mean IoU of Area_5_office_12: 0.5852
2021-03-22 17:13:30,382 - Model - INFO - Mean IoU of Area_5_office_28: 0.6707
2021-03-22 17:21:46,557 - Model - INFO - Mean IoU of Area_5_office_31: 0.6153
2021-03-22 17:26:49,769 - Model - INFO - Mean IoU of Area_5_office_26: 0.5207
2021-03-22 17:36:58,881 - Model - INFO - Mean IoU of Area_5_office_35: 0.7640
2021-03-22 23:59:28,625 - Model - INFO - Mean IoU of Area_5_hallway_2: 0.7175
2021-03-23 00:14:06,973 - Model - INFO - Mean IoU of Area_5_office_39: 0.6135
2021-03-23 00:52:12,769 - Model - INFO - Mean IoU of Area_5_office_24: 0.5573
2021-03-23 01:08:37,853 - Model - INFO - Mean IoU of Area_5_office_14: 0.6230
2021-03-23 01:11:29,081 - Model - INFO - Mean IoU of Area_5_storage_4: 0.6659
2021-03-23 01:28:36,731 - Model - INFO - Mean IoU of Area_5_hallway_6: 0.6781
2021-03-23 02:06:28,342 - Model - INFO - Mean IoU of Area_5_office_19: 0.7059
2021-03-23 02:14:30,956 - Model - INFO - Mean IoU of Area_5_office_42: 0.6049
2021-03-23 03:50:57,178 - Model - INFO - Mean IoU of Area_5_office_40: 0.4828
2021-03-23 03:57:10,565 - Model - INFO - Mean IoU of Area_5_office_4: 0.7075
2021-03-23 04:04:08,882 - Model - INFO - Mean IoU of Area_5_office_32: 0.6113
2021-03-23 05:12:29,614 - Model - INFO - Mean IoU of Area_5_office_37: 0.4237
2021-03-23 05:25:57,774 - Model - INFO - Mean IoU of Area_5_office_36: 0.6645
2021-03-23 05:30:57,288 - Model - INFO - Mean IoU of Area_5_office_6: 0.6999
2021-03-23 05:34:10,814 - Model - INFO - Mean IoU of Area_5_hallway_14: 0.7079
2021-03-23 06:04:14,884 - Model - INFO - Mean IoU of Area_5_office_29: 0.5924
2021-03-23 06:09:53,963 - Model - INFO - Mean IoU of Area_5_office_3: 0.7578
2021-03-23 06:15:24,359 - Model - INFO - Mean IoU of Area_5_hallway_4: 0.4651
2021-03-23 06:20:05,846 - Model - INFO - Mean IoU of Area_5_storage_1: 0.3533
2021-03-23 06:26:04,559 - Model - INFO - Mean IoU of Area_5_office_11: 0.6504
2021-03-23 06:32:53,686 - Model - INFO - Mean IoU of Area_5_office_8: 0.6385
2021-03-23 06:38:59,374 - Model - INFO - Mean IoU of Area_5_office_34: 0.6091
2021-03-23 07:29:16,087 - Model - INFO - Mean IoU of Area_5_office_21: 0.6285
2021-03-23 07:36:48,993 - Model - INFO - Mean IoU of Area_5_storage_2: 0.4328
2021-03-23 07:40:48,011 - Model - INFO - Mean IoU of Area_5_office_9: 0.7193
2021-03-23 07:42:18,540 - Model - INFO - Mean IoU of Area_5_office_30: 0.8101
2021-03-23 07:49:04,036 - Model - INFO - Mean IoU of Area_5_hallway_11: 0.8154
2021-03-23 08:41:39,690 - Model - INFO - Mean IoU of Area_5_conferenceRoom_2: 0.6049
2021-03-23 08:45:25,196 - Model - INFO - Mean IoU of Area_5_office_5: 0.8352
2021-03-23 08:52:35,233 - Model - INFO - Mean IoU of Area_5_hallway_12: 0.6916
2021-03-23 08:59:25,839 - Model - INFO - Mean IoU of Area_5_office_7: 0.6239
2021-03-23 09:01:18,197 - Model - INFO - Mean IoU of Area_5_hallway_8: 0.6584
2021-03-23 09:07:32,416 - Model - INFO - Mean IoU of Area_5_office_33: 0.6477
2021-03-23 09:29:47,860 - Model - INFO - Mean IoU of Area_5_hallway_9: 0.5687
2021-03-23 10:52:54,462 - Model - INFO - Mean IoU of Area_5_office_38: 0.3907
2021-03-23 11:24:48,653 - Model - INFO - Mean IoU of Area_5_hallway_10: 0.8178
2021-03-23 11:38:13,811 - Model - INFO - Mean IoU of Area_5_office_13: 0.6312
2021-03-23 11:45:05,688 - Model - INFO - Mean IoU of Area_5_WC_2: 0.7661
2021-03-23 11:53:47,220 - Model - INFO - Mean IoU of Area_5_office_23: 0.6462
2021-03-23 11:58:07,511 - Model - INFO - Mean IoU of Area_5_pantry_1: 0.7116
2021-03-23 12:06:09,280 - Model - INFO - Mean IoU of Area_5_office_22: 0.6502
2021-03-23 12:17:21,391 - Model - INFO - Mean IoU of Area_5_office_17: 0.5784
2021-03-23 12:26:08,789 - Model - INFO - Mean IoU of Area_5_WC_1: 0.7258
2021-03-23 12:34:21,005 - Model - INFO - Mean IoU of Area_5_office_2: 0.7176
2021-03-23 14:53:38,858 - Model - INFO - Mean IoU of Area_5_hallway_5: 0.5241
2021-03-23 15:09:10,854 - Model - INFO - Mean IoU of Area_5_conferenceRoom_1: 0.5500
2021-03-23 15:41:32,436 - Model - INFO - Mean IoU of Area_5_hallway_15: 0.5137
2021-03-23 15:49:09,586 - Model - INFO - Mean IoU of Area_5_office_16: 0.6031
2021-03-23 16:13:21,947 - Model - INFO - Mean IoU of Area_5_hallway_7: 0.7646
2021-03-23 16:35:25,771 - Model - INFO - Mean IoU of Area_5_hallway_3: 0.7395
2021-03-23 16:35:42,819 - Model - INFO - Mean IoU of Area_5_storage_3: 0.5660
2021-03-23 16:41:52,716 - Model - INFO - Mean IoU of Area_5_office_25: 0.6903
2021-03-23 16:46:08,458 - Model - INFO - Mean IoU of Area_5_office_20: 0.5930
2021-03-23 22:45:28,676 - Model - INFO - Mean IoU of Area_5_hallway_1: 0.3561
2021-03-23 22:45:35,270 - Model - INFO - ------- IoU --------
class ceiling , IoU: 0.894
class floor , IoU: 0.977
class wall , IoU: 0.754
class beam , IoU: 0.000
class column , IoU: 0.018
class window , IoU: 0.583
class door , IoU: 0.195
class table , IoU: 0.692
class chair , IoU: 0.790
class sofa , IoU: 0.462
class bookcase , IoU: 0.591
class board , IoU: 0.587
class clutter , IoU: 0.416

2021-03-23 22:45:35,271 - Model - INFO - eval point avg class IoU: 0.535313
2021-03-23 22:45:35,271 - Model - INFO - eval whole scene point avg class acc: 0.620289
2021-03-23 22:45:35,271 - Model - INFO - eval whole scene point accuracy: 0.829893
Loading

0 comments on commit c28e1b6

Please sign in to comment.