pytorch的grid_sample返回不正确的值
pytorch的grid_sample返回不正确的值
提问于 2021-06-25 20:54:12
我有一个3D矩阵：img[i, j, k] = i+j+k。
在我看来，如果我想要(1，2，3)的值，grid_sample应该返回6。但事实并非如此。
代码是：
import torch
from torch.nn import functional as F
import numpy as np
X, Y, Z = 10, 20, 30
img = np.zeros(shape=[X, Y, Z], dtype=np.float32)
for i in range(X):
for j in range(Y):
for k in range(Z):
img[i,j,k] = i+j+k
inp = torch.from_numpy(img).unsqueeze(0).unsqueeze(0)
grid = torch.from_numpy(np.array([[1, 2, 3]], dtype=np.float32)).unsqueeze(1).unsqueeze(1).unsqueeze(1)
grid[..., 0] /= (X-1)
grid[..., 1] /= (Y-1)
grid[..., 2] /= (Z-1)
grid = 2*grid - 1
outp = F.grid_sample(inp, grid=grid, mode='bilinear', align_corners=True)
print(outp)grid_sample返回6.15。我的代码有什么问题吗？
回答 1
Stack Overflow用户
发布于 2021-06-27 09:48:36
最后，我找到了解决方案。上面代码返回不正确的值的原因是torch.grid_sample接受(z，y，x)点。
因此，正确的代码应该是：
import torch
from torch.nn import functional as F
import numpy as np
X, Y, Z = 10, 20, 30
img = np.zeros(shape=[X, Y, Z], dtype=np.float32)
for i in range(X):
for j in range(Y):
for k in range(Z):
img[i,j,k] = i+j+k
inp = torch.from_numpy(img).unsqueeze(0).unsqueeze(0)
grid = torch.from_numpy(np.array([[1, 2, 3]], dtype=np.float32)).unsqueeze(1).unsqueeze(1).unsqueeze(1)
grid[..., 0] /= (X-1)
grid[..., 1] /= (Y-1)
grid[..., 2] /= (Z-1)
grid = 2*grid - 1
newgrid = grid.clone()
newgrid[..., 0] = grid[..., 2]
newgrid[..., 1] = grid[..., 1]
newgrid[..., 2] = grid[..., 0]
outp = F.grid_sample(inp, grid=newgrid, mode='bilinear', align_corners=True)
print(outp)页面原文内容由Stack Overflow提供。腾讯云小微IT领域专用引擎提供翻译支持
原文链接：
https://stackoverflow.com/questions/68131325
复制相关文章
相似问题
腾讯云开发者
