T-SNE降维可视化

代码并非最终版，需要一定基础，大部分写完了，最后调用没写。

from sklearn.manifold import TSNE
import matplotlib.pyplot as plt
from torchvision import models
import torch
import torch.nn as nn
from torch.utils.data import  DataLoader
import torchvision
from torch.nn import init
from torchvision import transforms
#测试数据集地址
val_path = "./RAF-DB/test"
#利用torch自带图片增强API，处理测试数据集
#可重写dataset函数，只读一个batch，无需全读
val_transform = transforms.Compose([
    transforms.Resize((224,224)),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.485,0.456,0.406],std=[0.229,0.224,0.225])
])
batch_size = 128
val_data = torchvision.datasets.ImageFolder(val_path, transform=val_transform)
data1_val = DataLoader(val_data, batch_size=batch_size, shuffle=True,drop_last=True)

#构建ResNet18模型
resnet = models.resnet18()
class ResNet18_(nn.Module):
    def __init__(self, num_class=7):
        super(ResNet18_, self).__init__()
        self.features = nn.Sequential(*list(resnet.children())[:-2])
        self.avgpool = nn.AdaptiveAvgPool2d(1)
        self.fc = nn.Linear(512, num_class)

    def forward(self, x):
        
        x = self.features(x)
        out = self.avgpool(x)
        out = torch.flatten(out,1)
        out = self.fc(out)
        return out

#模型实例化
model1 = ResNet18_()

#模型权重加载
model_path1 = "./OneModel/迁移学习/ResNet18/Base/qyxx-resnet18.pkl"
checkpoint1 = torch.load(model_path1,map_location='cpu')
model1.load_state_dict(checkpoint1['model'])
del checkpoint1

def evalute_(model1,val_loader):
    model1.eval()
    for batchidx, (x, label) in enumerate(val_loader):
        with torch.no_grad():
            y1 = model1(x)
            break
    y1 = y1.numpy()
    return y1,label

x1, y1 = evalute_(model1=model1,val_loader=data1_val)

#建立标签
tar = ['anger','disgust','fear','happy','neutral','sadness','surprise']

#构建可视化（x，y）
def T_SNE(pred):
    pred1 = TSNE(n_components=2, learning_rate='auto',init='pca').fit_transform(pred)
    #归一化
    pred1_min,pred_max = pred1.min(0), pred1.max(0)
    pred1 = (pred1 - pred1_min) / (pred_max - pred1_min)
    return pred1


#画图
def drawing(pred,label):
    plt.figure(figsize=(8, 8))
    for i in range(pred.shape[0]):
        plt.text(pred[i, 0], pred[i, 1], tar[label[i].item()], color=plt.cm.Set1(label[i]),\
        fontdict={'weight': 'bold', 'size': 9})
    plt.xticks([])
    plt.yticks([])
    plt.show()
    return 0

最终效果可视化