PyTorch学习笔记（十七）——完整的模型验证（测试，demo）套路

完整代码：

import torch
import torchvision
from PIL import Image
from torch import nnimage_path = "../imgs/dog.png"
image = Image.open(image_path)
print(image)# 因为png格式是四个通道，除了RGB三通道外，还有一个透明度通道
image = image.convert("RGB")
transform = torchvision.transforms.Compose([torchvision.transforms.Resize((32,32)),torchvision.transforms.ToTensor()])
image = transform(image)
print(image.shape)class MyNN(nn.Module):def __init__(self):super(MyNN, self).__init__()self.model = nn.Sequential(nn.Conv2d(3, 32, 5, 1, 2),nn.MaxPool2d(2),nn.Conv2d(32, 32, 5, 1, 2),nn.MaxPool2d(2),nn.Conv2d(32, 64, 5, 1, 2),nn.MaxPool2d(2),nn.Flatten(),nn.Linear(64 * 4 * 4, 64),nn.Linear(64, 10))def forward(self, x):x = self.model(x)return xmodel = torch.load("mynn_0.pth")
print(model)image = torch.reshape(image,(1,3,32,32))
model.eval()with torch.no_grad():output = model(image.cuda())
print(output)
print(output.argmax(1))

 采用GPU训练的模型，两种方法

（1）在CPU上加载，要从GPU映射到CPU，即把model = torch.load("mynn_9.pth")改为：

model = torch.load("mynn_9.pth",map_location=torch.device('cpu'))

（2）将image转到GPU中，即把output = model(image)改为：

output = model(image.cuda())

 预测错误的原因可能是训练次数不够多

 改成：

model = torch.load("mynn_9.pth")