-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathscripts.sh
97 lines (62 loc) · 4.75 KB
/
scripts.sh
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
### Train (we train without any rotation, then test with random rotation in 3D space)
## On ModelNet40
## Based on PointNet, full-precision
python main_cls_pointnet.py --model=svnet --data-dir data --save-dir result/train --rot aligned --rot-test so3
## Based on PointNet, binary
python main_cls_pointnet.py --model=svnet --data-dir data --save-dir result/train --rot aligned --rot-test so3 --binary --wd 0
## Based on DGCNN, full-precision
python main_cls_dgcnn.py --model=svnet --data-dir data --save-dir result/train --rot aligned --rot-test so3
## Based on DGCNN, binary
python main_cls_dgcnn.py --model=svnet --data-dir data --save-dir result/train --rot aligned --rot-test so3 --binary --wd 0
## On ShapeNet
## Based on PointNet, full-precision
python main_partseg_pointnet.py --model=svnet --data-dir data --save-dir result/train --rot aligned --rot-test so3
## Based on PointNet, binary
python main_partseg_pointnet.py --model=svnet --data-dir data --save-dir result/train --rot aligned --rot-test so3 --binary --wd 0
## Based on DGCNN, full-precision
python main_partseg_dgcnn.py --model=svnet --data-dir data --save-dir result/train --rot aligned --rot-test so3
## Based on DGCNN, binary
python main_partseg_dgcnn.py --model=svnet --data-dir data --save-dir result/train --rot aligned --rot-test so3 --binary --wd 0
## On ScanObjectNN
## Based on DGCNN, full-precision
python main_cls_dgcnn.py --dataset scanobjectnn --model=svnet --data-dir /data/scanobjectnn --save-dir result/train --rot aligned --rot-test so3
## Based on DGCNN, binary
python main_cls_dgcnn.py --dataset scanobjectnn --model=svnet --data-dir /data/scanobjectnn --save-dir result/train --rot aligned --rot-test so3 --binary --wd 0
### Evaluation (with random rotation in 3D space)
## On ModelNet40
## Based on PointNet, full-precision
python main_cls_pointnet.py --model=svnet --batch-size 16 --data-dir data --save-dir result/test --rot-test so3 --test checkpoints/sv_pointnet_fp_modelnet40.pth
## Based on PointNet, binary
python main_cls_pointnet.py --model=svnet --batch-size 16 --data-dir data --save-dir result/test --rot-test so3 --test checkpoints/sv_pointnet_binary_modelnet40.pth --binary
## Based on DGCNN, full-precision
python main_cls_dgcnn.py --model=svnet --batch-size 16 --data-dir data --save-dir result/test --rot-test so3 --test checkpoints/sv_dgcnn_fp_modelnet40.pth
## Based on DGCNN, binary
python main_cls_dgcnn.py --model=svnet --batch-size 16 --data-dir data --save-dir result/test --rot-test so3 --test checkpoints/sv_dgcnn_binary_modelnet40.pth --binary
# with knowledge distillation
python main_cls_dgcnn.py --model=svnet --batch-size 16 --data-dir data --save-dir result/test --rot-test so3 --test checkpoints/sv_dgcnn_binary_kd_modelnet40.pth --binary
## On ShapeNet
## Based on PointNet, full-precision
python main_partseg_pointnet.py --model=svnet --batch-size 4 --data-dir data --save-dir result/test --rot-test so3 --test checkpoints/sv_pointnet_fp_shapenet.pth
## Based on PointNet, binary
python main_partseg_pointnet.py --model=svnet --batch-size 4 --data-dir data --save-dir result/test --rot-test so3 --test checkpoints/sv_pointnet_binary_shapenet.pth --binary
## Based on DGCNN, full-precision
python main_partseg_dgcnn.py --model=svnet --batch-size 4 --data-dir data --save-dir result/test --rot-test so3 --test checkpoints/sv_dgcnn_fp_shapenet.pth
## Based on DGCNN, binary
python main_partseg_dgcnn.py --model=svnet --batch-size 4 --data-dir data --save-dir result/test --rot-test so3 --test checkpoints/sv_dgcnn_binary_shapenet.pth --binary
# with knowledge distillation
python main_partseg_dgcnn.py --model=svnet --batch-size 4 --data-dir data --save-dir result/test --rot-test so3 --test checkpoints/sv_dgcnn_binary_kd_shapenet.pth --binary
## On ScanObjectNN
## Based on DGCNN, full-precision
python main_cls_dgcnn.py --dataset scanobjectnn --model=svnet --batch-size 16 --data-dir /data2/zhuo/dataset/pointcloud/scanobjectnn --save-dir zhuo_test/test --rot-test so3 --test checkpoints/sv_dgcnn_fp_scanobjectnn.pth
## Based on DGCNN, binary
python main_cls_dgcnn.py --dataset scanobjectnn --model=svnet --batch-size 16 --data-dir /data2/zhuo/dataset/pointcloud/scanobjectnn --save-dir zhuo_test/test --rot-test so3 --test checkpoints/sv_dgcnn_binary_scanobjectnn.pth --binary
# with knowledge distillation
python main_cls_dgcnn.py --dataset scanobjectnn --model=svnet --batch-size 16 --data-dir /data2/zhuo/dataset/pointcloud/scanobjectnn --save-dir zhuo_test/test --rot-test so3 --test checkpoints/sv_dgcnn_binary_kd_scanobjectnn.pth --binary
### Check model size and FLOPs/BOPs
python params_macs/sv_pointnet.py
python params_macs/sv_dgcnn.py
python params_macs/vn_pointnet.py
python params_macs/vn_dgcnn.py
python params_macs/pointnet.py
python params_macs/dgcnn.py
python params_macs/bipointnet.py