The size of tensor a (4) must match the size of tensor b (3) at non-singleton dimension 0 维度不匹配

在导入一个png文件的时候，发现报这个错，源码如下，经过查询找到了两种解决方案。

from PIL import Image
from torchvision import transforms
from torch.utils.tensorboard import SummaryWriter
import torchwriter = SummaryWriter("logs")
img = Image.open("C:/Users/Mxx/PycharmProjects/pythonProject4/images_set/GOGOGOGO.png")
print(img)
#img = img.convert('RGB')
print(img)trans_totensor = transforms.ToTensor() #首先创建一个对象，来使用ToTensor的方法
img_tensor = trans_totensor(img)
writer.add_image("ToTensor",img_tensor)#img = img.convert('RGB')
#Normalize
print(img_tensor[0][0][0])
print("---------")
print(img_tensor.shape)
trans_norm = transforms.Normalize([0.5,0.5,0.5],[0.5,0.5,0.5])
img_norm = trans_norm(img_tensor)
print(img_norm[0][0][0])
writer.add_image("Normalize",img_norm)
writer.close()

1.因为png是四个通道，所以需要在Normalize的时候加一个维度。

trans_norm = transforms.Normalize([0.5,0.5,0.5,0.5],[0.5,0.5,0.5,0.5])

2.讲png转成rgb这样只需要做三个维度的标准化。

img = img.convert('RGB')

将png和RGB以及标准化后的png和RGB进行输出，发现原本的png和RGB差不多。

标准化后的相比于标准化前是变的模糊了，但是标准化后的png和RGB是差不多的。

更改后的test代码如下

from PIL import Image
from torchvision import transforms
from torch.utils.tensorboard import SummaryWriter
import torchwriter = SummaryWriter("logs")
img = Image.open("C:/Users/Mxx/PycharmProjects/pythonProject4/images_set/GOGOGOGO.png")
print(img)
img_RGB = img.convert('RGB')trans_totensor = transforms.ToTensor() #首先创建一个对象，来使用ToTensor的方法
img_tensor = trans_totensor(img)
#rgb 的 tensor
imgRGB_tensor = trans_totensor(img_RGB)writer.add_image("ToTensor",img_tensor)writer.add_image("RGB-ToTensor",imgRGB_tensor)
#img = img.convert('RGB')
#Normalizetrans_norm_png = transforms.Normalize([0.5,0.5,0.5,0.5],[0.5,0.5,0.5,0.5])
img_norm = trans_norm_png(img_tensor)trans_norm_RGB = transforms.Normalize([0.5,0.5,0.5],[0.5,0.5,0.5])
imgRGB_norm = trans_norm_RGB(imgRGB_tensor)writer.add_image("PNG Normalize",img_norm)
writer.add_image("RGB Normalize",imgRGB_norm)
writer.close()

The size of tensor a (4) must match the size of tensor b (3) at non-singleton dimension 0 维度不匹配相关推荐

The size of tensor a (4) must match the size of tensor b (3) at non-singletonThe size of
报错: RuntimeError: The size of tensor a (4) must match the size of tensor b (3) at non-singleton 解决: ...
报错解决——RuntimeError: The size of tensor a (4) must match the size of tensor b (3) at non-singleton
问题描述 RuntimeError: The size of tensor a (4) must match the size of tensor b (3) at non-singleton dim ...
CUDA编译器nvcc可能遇到：error: asm operand type size(8) does not match type/size implied by constraint ‘r‘
0. 问题描述该问题产生于我在ptcharm中试图用cupy加速numpy计算,进行一些简单但是繁琐的数据处理,这个问题和cupy关系不大,cupy只要照着nvcc -v给出的版本号对应着下载就ok ...
RuntimeError: invalid argument 0: Sizes of tensors must match except in dimension 0—问题分析
声明: 这里仅仅提供一个思路,暂时没有绝对正确通用的解决方案作者:{ 墨理 } 感谢通过私信.评论交流,协力探讨解决问题的同学 UP .Penser 文章目录探讨的 pytorch 报错信息如下 ...
PyTorch 笔记（11）— Tensor内部存储结构（头信息区 Tensor，存储区 Storage）
1. Tensor 内部存储结构 tensor 数据结构如下图所示,tensor 分为头信息区(Tensor)和存储区 (Storage),信息区主要保存着 Tensor 的形状(size).步长(s ...
python中tensor与variable_NLP实战篇之tf2中tensor、variable、gradient、ops
本文是基于tensorflow2.2.0版本,介绍了tf中变量.张量的概念,tf中梯度的计算方式和tensor相关的操作. 实战系列篇章中主要会分享,解决实际问题时的过程.遇到的问题或者使用的工具等等 ...
pytorch dataload 出错：: invalid argument 0: Sizes of tensors must match except in dimension 0
错误信息:"RuntimeError: invalid argument 0: Sizes of tensors must match except in dimension 0. Got ...
Pytorch_Geometric（PyG）使用DataLoader报错RuntimeError: Sizes of tensors must match except in dimension 0.
使用Pytorch_Geometric(PyG)时构建DataLoader,从DataLoader获取样本Batch时报错:RuntimeError: Sizes of tensors must ma ...
Pytorch踩坑: RuntimeError: invalid argument 0: Sizes of tensors must match except in dimension 0.
报错信息: RuntimeError: invalid argument 0: Sizes of tensors must match except in dimension 0. Got 186 a ...

The size of tensor a (4) must match the size of tensor b (3) at non-singleton dimension 0 维度不匹配

The size of tensor a (4) must match the size of tensor b (3) at non-singleton dimension 0 维度不匹配相关推荐

最新文章

热门文章