lss
1. create_frustum
def create_frustum(self):
# make grid in image plane
ogfH, ogfW = self.data_aug_conf['final_dim'] #ogfH:128 ogfW:352
fH, fW = ogfH // self.downsample, ogfW // self.downsample #downsample16 fH:8 fW:22
#shape tmp(41) 4,5,6,...,44 #grid_conf['dbound'] [4,45,1]
tmp = torch.arange(*self.grid_conf['dbound'], dtype=torch.float)
#ds [41, 8, 22]
ds = torch.arange(*self.grid_conf['dbound'], dtype=torch.float).view(-1, 1, 1).expand(-1, fH, fW)
D, _, _ = ds.shape #D=41
#tmp_xs[22] ogfW:352 fW:22 fH:8 fW:22 (351/21=16.7143)
tmp_xs = torch.linspace(0, ogfW - 1, fW, dtype=torch.float)
#tmp_ys[8] ogfH:128 fH:8 (127/7=18.1429)
tmp_ys = torch.linspace(0, ogfH - 1, fH, dtype=torch.float)
# xs, ys [41, 8, 22]
xs = torch.linspace(0, ogfW - 1, fW, dtype=torch.float).view(1, 1, fW).expand(D, fH, fW)
ys = torch.linspace(0, ogfH - 1, fH, dtype=torch.float).view(1, fH, 1).expand(D, fH, fW)
# D x H x W x 3 [41, 8, 22, 3]
frustum = torch.stack((xs, ys, ds), -1)
return nn.Parameter(frustum, requires_grad=False)
test_create_frustum.py
import torch
downsample = 3
ogfH = 21
ogfW = 36
#fH=7 fW=12
fH, fW = ogfH // downsample, ogfW // downsample
# #[5, 10]
# tmp0 = torch.arange(*[5, 15, 5], dtype=torch.float)
#ds[2,7,12]
ds = torch.arange(*[5, 15, 5], dtype=torch.float).view(-1, 1, 1).expand(-1, fH, fW)
#D=2
D, _, _ = ds.shape
#tmp_xs[12] ogfW:36 fW:12 fH:7 (35/11=3.1818)
tmp_xs = torch.linspace(0, ogfW - 1, fW, dtype=torch.float)
#tmp_ys[7] ogfH:21 fH:7 (20/6=3.333)
tmp_ys = torch.linspace(0, ogfH - 1, fH, dtype=torch.float)
#[1, 1, 12]
tmp_xs_view = torch.linspace(0, ogfW - 1, fW, dtype=torch.float).view(1, 1, fW)
#[1, 7, 1]
tmp_ys_view = torch.linspace(0, ogfH - 1, fH, dtype=torch.float).view(1, fH, 1)
# xs, ys [2, 7, 12]
xs = torch.linspace(0, ogfW - 1, fW, dtype=torch.float).view(1, 1, fW).expand(D, fH, fW)
ys = torch.linspace(0, ogfH - 1, fH, dtype=torch.float).view(1, fH, 1).expand(D, fH, fW)
# D x H x W x 3 [2, 7, 12, 3]
frustum = torch.stack((xs, ys, ds), -1)
print("==============>>>>>>>>>>>>>>>tmp_xs_view.shape=", tmp_xs_view.shape)
print(tmp_xs_view)
print("==============>>>>>>>>>>>>>>>xs.shape=", xs.shape)
print(xs)
print("==============>>>>>>>>>>>>>>>tmp_ys_view.shape=", tmp_ys_view.shape)
print(tmp_ys_view)
print("==============>>>>>>>>>>>>>>>ys.shape=", ys.shape)
print(ys)
print("==============>>>>>>>>>>>>>>>ds.shape=", ds.shape)
print(ds)
print("==============>>>>>>>>>>>>>>>frustum.shape=", frustum.shape)
print(frustum)
/media/algo/data_1/software/anconda_install/envs/pytorch1.7.0_general/bin/python3 /media/algo/data_1/project_others/0000paper/lss/project/lift-splat-shoot-master/0000/create_frustum.py
==============>>>>>>>>>>>>>>>tmp_xs_view.shape= torch.Size([1, 1, 12])
tensor([[[ 0.0000, 3.1818, 6.3636, 9.5455, 12.7273, 15.9091, 19.0909,
22.2727, 25.4545, 28.6364, 31.8182, 35.0000]]])
==============>>>>>>>>>>>>>>>xs.shape= torch.Size([2, 7, 12])
tensor([[[ 0.0000, 3.1818, 6.3636, 9.5455, 12.7273, 15.9091, 19.0909,
22.2727, 25.4545, 28.6364, 31.8182, 35.0000],
[ 0.0000, 3.1818, 6.3636, 9.5455, 12.7273, 15.9091, 19.0909,
22.2727, 25.4545, 28.6364, 31.8182, 35.0000],
[ 0.0000, 3.1818, 6.3636, 9.5455, 12.7273, 15.9091, 19.0909,
22.2727, 25.4545, 28.6364, 31.8182, 35.0000],
[ 0.0000, 3.1818, 6.3636, 9.5455, 12.7273, 15.9091, 19.0909,
22.2727, 25.4545, 28.6364, 31.8182, 35.0000],
[ 0.0000, 3.1818, 6.3636, 9.5455, 12.7273, 15.9091, 19.0909,
22.2727, 25.4545, 28.6364, 31.8182, 35.0000],
[ 0.0000, 3.1818, 6.3636, 9.5455, 12.7273, 15.9091, 19.0909,
22.2727, 25.4545, 28.6364, 31.8182, 35.0000],
[ 0.0000, 3.1818, 6.3636, 9.5455, 12.7273, 15.9091, 19.0909,
22.2727, 25.4545, 28.6364, 31.8182, 35.0000]],
[[ 0.0000, 3.1818, 6.3636, 9.5455, 12.7273, 15.9091, 19.0909,
22.2727, 25.4545, 28.6364, 31.8182, 35.0000],
[ 0.0000, 3.1818, 6.3636, 9.5455, 12.7273, 15.9091, 19.0909,
22.2727, 25.4545, 28.6364, 31.8182, 35.0000],
[ 0.0000, 3.1818, 6.3636, 9.5455, 12.7273, 15.9091, 19.0909,
22.2727, 25.4545, 28.6364, 31.8182, 35.0000],
[ 0.0000, 3.1818, 6.3636, 9.5455, 12.7273, 15.9091, 19.0909,
22.2727, 25.4545, 28.6364, 31.8182, 35.0000],
[ 0.0000, 3.1818, 6.3636, 9.5455, 12.7273, 15.9091, 19.0909,
22.2727, 25.4545, 28.6364, 31.8182, 35.0000],
[ 0.0000, 3.1818, 6.3636, 9.5455, 12.7273, 15.9091, 19.0909,
22.2727, 25.4545, 28.6364, 31.8182, 35.0000],
[ 0.0000, 3.1818, 6.3636, 9.5455, 12.7273, 15.9091, 19.0909,
22.2727, 25.4545, 28.6364, 31.8182, 35.0000]]])
==============>>>>>>>>>>>>>>>tmp_ys_view.shape= torch.Size([1, 7, 1])
tensor([[[ 0.0000],
[ 3.3333],
[ 6.6667],
[10.0000],
[13.3333],
[16.6667],
[20.0000]]])
==============>>>>>>>>>>>>>>>ys.shape= torch.Size([2, 7, 12])
tensor([[[ 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
0.0000, 0.0000, 0.0000, 0.0000, 0.0000],
[ 3.3333, 3.3333, 3.3333, 3.3333, 3.3333, 3.3333, 3.3333,
3.3333, 3.3333, 3.3333, 3.3333, 3.3333],
[ 6.6667, 6.6667, 6.6667, 6.6667, 6.6667, 6.6667, 6.6667,
6.6667, 6.6667, 6.6667, 6.6667, 6.6667],
[10.0000, 10.0000, 10.0000, 10.0000, 10.0000, 10.0000, 10.0000,
10.0000, 10.0000, 10.0000, 10.0000, 10.0000],
[13.3333, 13.3333, 13.3333, 13.3333, 13.3333, 13.3333, 13.3333,
13.3333, 13.3333, 13.3333, 13.3333, 13.3333],
[16.6667, 16.6667, 16.6667, 16.6667, 16.6667, 16.6667, 16.6667,
16.6667, 16.6667, 16.6667, 16.6667, 16.6667],
[20.0000, 20.0000, 20.0000, 20.0000, 20.0000, 20.0000, 20.0000,
20.0000, 20.0000, 20.0000, 20.0000, 20.0000]],
[[ 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000, 0.0000,
0.0000, 0.0000, 0.0000, 0.0000, 0.0000],
[ 3.3333, 3.3333, 3.3333, 3.3333, 3.3333, 3.3333, 3.3333,
3.3333, 3.3333, 3.3333, 3.3333, 3.3333],
[ 6.6667, 6.6667, 6.6667, 6.6667, 6.6667, 6.6667, 6.6667,
6.6667, 6.6667, 6.6667, 6.6667, 6.6667],
[10.0000, 10.0000, 10.0000, 10.0000, 10.0000, 10.0000, 10.0000,
10.0000, 10.0000, 10.0000, 10.0000, 10.0000],
[13.3333, 13.3333, 13.3333, 13.3333, 13.3333, 13.3333, 13.3333,
13.3333, 13.3333, 13.3333, 13.3333, 13.3333],
[16.6667, 16.6667, 16.6667, 16.6667, 16.6667, 16.6667, 16.6667,
16.6667, 16.6667, 16.6667, 16.6667, 16.6667],
[20.0000, 20.0000, 20.0000, 20.0000, 20.0000, 20.0000, 20.0000,
20.0000, 20.0000, 20.0000, 20.0000, 20.0000]]])
==============>>>>>>>>>>>>>>>ds.shape= torch.Size([2, 7, 12])
tensor([[[ 5., 5., 5., 5., 5., 5., 5., 5., 5., 5., 5., 5.],
[ 5., 5., 5., 5., 5., 5., 5., 5., 5., 5., 5., 5.],
[ 5., 5., 5., 5., 5., 5., 5., 5., 5., 5., 5., 5.],
[ 5., 5., 5., 5., 5., 5., 5., 5., 5., 5., 5., 5.],
[ 5., 5., 5., 5., 5., 5., 5., 5., 5., 5., 5., 5.],
[ 5., 5., 5., 5., 5., 5., 5., 5., 5., 5., 5., 5.],
[ 5., 5., 5., 5., 5., 5., 5., 5., 5., 5., 5., 5.]],
[[10., 10., 10., 10., 10., 10., 10., 10., 10., 10., 10., 10.],
[10., 10., 10., 10., 10., 10., 10., 10., 10., 10., 10., 10.],
[10., 10., 10., 10., 10., 10., 10., 10., 10., 10., 10., 10.],
[10., 10., 10., 10., 10., 10., 10., 10., 10., 10., 10., 10.],
[10., 10., 10., 10., 10., 10., 10., 10., 10., 10., 10., 10.],
[10., 10., 10., 10., 10., 10., 10., 10., 10., 10., 10., 10.],
[10., 10., 10., 10., 10., 10., 10., 10., 10., 10., 10., 10.]]])
==============>>>>>>>>>>>>>>>frustum.shape= torch.Size([2, 7, 12, 3])
tensor([[[[ 0.0000, 0.0000, 5.0000],
[ 3.1818, 0.0000, 5.0000],
[ 6.3636, 0.0000, 5.0000],
[ 9.5455, 0.0000, 5.0000],
[12.7273, 0.0000, 5.0000],
[15.9091, 0.0000, 5.0000],
[19.0909, 0.0000, 5.0000],
[22.2727, 0.0000, 5.0000],
[25.4545, 0.0000, 5.0000],
[28.6364, 0.0000, 5.0000],
[31.8182, 0.0000, 5.0000],
[35.0000, 0.0000, 5.0000]],
[[ 0.0000, 3.3333, 5.0000],
[ 3.1818, 3.3333, 5.0000],
[ 6.3636, 3.3333, 5.0000],
[ 9.5455, 3.3333, 5.0000],
[12.7273, 3.3333, 5.0000],
[15.9091, 3.3333, 5.0000],
[19.0909, 3.3333, 5.0000],
[22.2727, 3.3333, 5.0000],
[25.4545, 3.3333, 5.0000],
[28.6364, 3.3333, 5.0000],
[31.8182, 3.3333, 5.0000],
[35.0000, 3.3333, 5.0000]],
[[ 0.0000, 6.6667, 5.0000],
[ 3.1818, 6.6667, 5.0000],
[ 6.3636, 6.6667, 5.0000],
[ 9.5455, 6.6667, 5.0000],
[12.7273, 6.6667, 5.0000],
[15.9091, 6.6667, 5.0000],
[19.0909, 6.6667, 5.0000],
[22.2727, 6.6667, 5.0000],
[25.4545, 6.6667, 5.0000],
[28.6364, 6.6667, 5.0000],
[31.8182, 6.6667, 5.0000],
[35.0000, 6.6667, 5.0000]],
[[ 0.0000, 10.0000, 5.0000],
[ 3.1818, 10.0000, 5.0000],
[ 6.3636, 10.0000, 5.0000],
[ 9.5455, 10.0000, 5.0000],
[12.7273, 10.0000, 5.0000],
[15.9091, 10.0000, 5.0000],
[19.0909, 10.0000, 5.0000],
[22.2727, 10.0000, 5.0000],
[25.4545, 10.0000, 5.0000],
[28.6364, 10.0000, 5.0000],
[31.8182, 10.0000, 5.0000],
[35.0000, 10.0000, 5.0000]],
[[ 0.0000, 13.3333, 5.0000],
[ 3.1818, 13.3333, 5.0000],
[ 6.3636, 13.3333, 5.0000],
[ 9.5455, 13.3333, 5.0000],
[12.7273, 13.3333, 5.0000],
[15.9091, 13.3333, 5.0000],
[19.0909, 13.3333, 5.0000],
[22.2727, 13.3333, 5.0000],
[25.4545, 13.3333, 5.0000],
[28.6364, 13.3333, 5.0000],
[31.8182, 13.3333, 5.0000],
[35.0000, 13.3333, 5.0000]],
[[ 0.0000, 16.6667, 5.0000],
[ 3.1818, 16.6667, 5.0000],
[ 6.3636, 16.6667, 5.0000],
[ 9.5455, 16.6667, 5.0000],
[12.7273, 16.6667, 5.0000],
[15.9091, 16.6667, 5.0000],
[19.0909, 16.6667, 5.0000],
[22.2727, 16.6667, 5.0000],
[25.4545, 16.6667, 5.0000],
[28.6364, 16.6667, 5.0000],
[31.8182, 16.6667, 5.0000],
[35.0000, 16.6667, 5.0000]],
[[ 0.0000, 20.0000, 5.0000],
[ 3.1818, 20.0000, 5.0000],
[ 6.3636, 20.0000, 5.0000],
[ 9.5455, 20.0000, 5.0000],
[12.7273, 20.0000, 5.0000],
[15.9091, 20.0000, 5.0000],
[19.0909, 20.0000, 5.0000],
[22.2727, 20.0000, 5.0000],
[25.4545, 20.0000, 5.0000],
[28.6364, 20.0000, 5.0000],
[31.8182, 20.0000, 5.0000],
[35.0000, 20.0000, 5.0000]]],
[[[ 0.0000, 0.0000, 10.0000],
[ 3.1818, 0.0000, 10.0000],
[ 6.3636, 0.0000, 10.0000],
[ 9.5455, 0.0000, 10.0000],
[12.7273, 0.0000, 10.0000],
[15.9091, 0.0000, 10.0000],
[19.0909, 0.0000, 10.0000],
[22.2727, 0.0000, 10.0000],
[25.4545, 0.0000, 10.0000],
[28.6364, 0.0000, 10.0000],
[31.8182, 0.0000, 10.0000],
[35.0000, 0.0000, 10.0000]],
[[ 0.0000, 3.3333, 10.0000],
[ 3.1818, 3.3333, 10.0000],
[ 6.3636, 3.3333, 10.0000],
[ 9.5455, 3.3333, 10.0000],
[12.7273, 3.3333, 10.0000],
[15.9091, 3.3333, 10.0000],
[19.0909, 3.3333, 10.0000],
[22.2727, 3.3333, 10.0000],
[25.4545, 3.3333, 10.0000],
[28.6364, 3.3333, 10.0000],
[31.8182, 3.3333, 10.0000],
[35.0000, 3.3333, 10.0000]],
[[ 0.0000, 6.6667, 10.0000],
[ 3.1818, 6.6667, 10.0000],
[ 6.3636, 6.6667, 10.0000],
[ 9.5455, 6.6667, 10.0000],
[12.7273, 6.6667, 10.0000],
[15.9091, 6.6667, 10.0000],
[19.0909, 6.6667, 10.0000],
[22.2727, 6.6667, 10.0000],
[25.4545, 6.6667, 10.0000],
[28.6364, 6.6667, 10.0000],
[31.8182, 6.6667, 10.0000],
[35.0000, 6.6667, 10.0000]],
[[ 0.0000, 10.0000, 10.0000],
[ 3.1818, 10.0000, 10.0000],
[ 6.3636, 10.0000, 10.0000],
[ 9.5455, 10.0000, 10.0000],
[12.7273, 10.0000, 10.0000],
[15.9091, 10.0000, 10.0000],
[19.0909, 10.0000, 10.0000],
[22.2727, 10.0000, 10.0000],
[25.4545, 10.0000, 10.0000],
[28.6364, 10.0000, 10.0000],
[31.8182, 10.0000, 10.0000],
[35.0000, 10.0000, 10.0000]],
[[ 0.0000, 13.3333, 10.0000],
[ 3.1818, 13.3333, 10.0000],
[ 6.3636, 13.3333, 10.0000],
[ 9.5455, 13.3333, 10.0000],
[12.7273, 13.3333, 10.0000],
[15.9091, 13.3333, 10.0000],
[19.0909, 13.3333, 10.0000],
[22.2727, 13.3333, 10.0000],
[25.4545, 13.3333, 10.0000],
[28.6364, 13.3333, 10.0000],
[31.8182, 13.3333, 10.0000],
[35.0000, 13.3333, 10.0000]],
[[ 0.0000, 16.6667, 10.0000],
[ 3.1818, 16.6667, 10.0000],
[ 6.3636, 16.6667, 10.0000],
[ 9.5455, 16.6667, 10.0000],
[12.7273, 16.6667, 10.0000],
[15.9091, 16.6667, 10.0000],
[19.0909, 16.6667, 10.0000],
[22.2727, 16.6667, 10.0000],
[25.4545, 16.6667, 10.0000],
[28.6364, 16.6667, 10.0000],
[31.8182, 16.6667, 10.0000],
[35.0000, 16.6667, 10.0000]],
[[ 0.0000, 20.0000, 10.0000],
[ 3.1818, 20.0000, 10.0000],
[ 6.3636, 20.0000, 10.0000],
[ 9.5455, 20.0000, 10.0000],
[12.7273, 20.0000, 10.0000],
[15.9091, 20.0000, 10.0000],
[19.0909, 20.0000, 10.0000],
[22.2727, 20.0000, 10.0000],
[25.4545, 20.0000, 10.0000],
[28.6364, 20.0000, 10.0000],
[31.8182, 20.0000, 10.0000],
[35.0000, 20.0000, 10.0000]]]])
Process finished with exit code 0
2. 测试expand
import torch
a = torch.Tensor([[1, 2, 3],
[4, 5, 6]])
print(a.shape)
e0 = a.view(2, 3, 1)
e00 = e0.expand(2, 3, 5)
print(e0)
print()
print(e00)
print("=====================================")
e0 = a.view(2, 1, 3)
e00 = e0.expand(2, 5, 3)
print(e0)
print()
print(e00)
torch.Size([2, 3])
tensor([[[1.],
[2.],
[3.]],
[[4.],
[5.],
[6.]]])
tensor([[[1., 1., 1., 1., 1.],
[2., 2., 2., 2., 2.],
[3., 3., 3., 3., 3.]],
[[4., 4., 4., 4., 4.],
[5., 5., 5., 5., 5.],
[6., 6., 6., 6., 6.]]])
=====================================
tensor([[[1., 2., 3.]],
[[4., 5., 6.]]])
tensor([[[1., 2., 3.],
[1., 2., 3.],
[1., 2., 3.],
[1., 2., 3.],
[1., 2., 3.]],
[[4., 5., 6.],
[4., 5., 6.],
[4., 5., 6.],
[4., 5., 6.],
[4., 5., 6.]]])
Process finished with exit code 0
2.1
import torch
t1 = torch.Tensor([ 8, 9])
t2 = t1.view(-1, 1)
t22 = t2.expand(-1, 3)
t3 = t1.view(-1, 1, 1)
t33 = t3.expand(-1, 3, 4)
print("t1.shape=", t1.shape)
print(t1)
print("t2.shape=", t2.shape)
print(t2)
print("t22.shape=", t22.shape)
print(t22)
print("t3.shape=", t3.shape)
print(t3)
print("t33.shape=", t33.shape)
print(t33)
/media/algo/data_1/software/anconda_install/envs/pytorch1.7.0_general/bin/python3 /media/algo/data_1/project_others/0000paper/lss/project/lift-splat-shoot-master/0000/expand_test.py
t1.shape= torch.Size([2])
tensor([8., 9.])
t2.shape= torch.Size([2, 1])
tensor([[8.],
[9.]])
t22.shape= torch.Size([2, 3])
tensor([[8., 8., 8.],
[9., 9., 9.]])
t3.shape= torch.Size([2, 1, 1])
tensor([[[8.]],
[[9.]]])
t33.shape= torch.Size([2, 3, 4])
tensor([[[8., 8., 8., 8.],
[8., 8., 8., 8.],
[8., 8., 8., 8.]],
[[9., 9., 9., 9.],
[9., 9., 9., 9.],
[9., 9., 9., 9.]]])
Process finished with exit code 0
3. points = self.frustum - post_trans.view(B, N, 1, 1, 1, 3)
B, N, _ = trans.shape
#self.frustum [41, 8, 22, 3] [D, H, W, 3]
# undo post-transformation
# B x N x D x H x W x 3 [41, 8, 22, 3] - [2, 5, 1, 1, 1, 3]
points = self.frustum - post_trans.view(B, N, 1, 1, 1, 3)
points = torch.inverse(post_rots).view(B, N, 1, 1, 1, 3, 3).matmul(points.unsqueeze(-1))
测试demo
import torch
a = torch.randint(1, 100, (2, 3))
b = torch.randint(1, 100, (3, 1, 1, 3))
print(a)
print(b)
c = a - b
print(c)
print(c.shape)
tensor([[37, 8, 68],
[58, 9, 5]])
tensor([[[[76, 1, 58]]],
[[[61, 81, 75]]],
[[[18, 9, 98]]]])
tensor([[[[-39, 7, 10],
[-18, 8, -53]]],
[[[-24, -73, -7],
[ -3, -72, -70]]],
[[[ 19, -1, -30],
[ 40, 0, -93]]]])
torch.Size([3, 1, 2, 3])
Process finished with exit code 0
好记性不如烂键盘---点滴、积累、进步！
