使用vgg19进行微表情分类

介绍：通过pytorch提供的预训练模型进行迁移学习，进行微表情分类。先说一下数据集：

Kaggle的另一项挑战，Facial Expression Recognition Challenge（FER）（面部表情识别挑战）的目标是仅从他们的面部照片就能正确的识别出一个人正在经历的面部表情。在FER中提供了35888张图像，目标是将给定的面部表情分类到下述的7种类别中：

1. Angry

2. Disgust (sometimes grouped in with “Fear” due to class imbalance)

3. Fear

4. Happy

5. Sad

6. Surprise

7. Neutral

下载地址：https://pan.baidu.com/s/1haTjyVqxi8gVC0Xd873Qxw

训练后精度大概有0.6吧。相对来说低一些，以后有机会再优化或者换模型试一试吧。   

贴上代码：

   

import torch.nn as nn
from torch.utils import data
from torchvision import transforms
from PIL import Image
from torchvision import models as MD
import torch
import torchvision.datasets as datasets


def  Make_data(path):
    img=datasets.ImageFolder(path,
                         tensform=transforms.Compose([transforms.Scale([48, 48]), transforms.ToTensor()]))
    imgs_load=data.DataLoader(img,batch_size=100,shuffle=True)
    return imgs_load


def train(dada_loader):
    model = MD.vgg19(pretrained=False)
    model.load_state_dict(torch.load("../models/???.pth"))
    num_input = model.classifier[6].in_features
    feature_model = list(model.classifier.children())
    feature_model.pop()
    feature_model.append(nn.Linear(num_input, 7))
    model.classifier = nn.Sequential(*feature_model)
    model = model.cuda()
    critersion = nn.CrossEntropyLoss()
    opt = torch.optim.SGD(model.parameters(), lr=0.001)

    for ench in range(200):
        sum = 0
        for i, data in enumerate(dada_loader):
            img, targe = data[1]
            targe = targe.cuda()
            img = img.cuda()
            output = model(img)
            loss = critersion(output, targe)
            opt.zero_grad()
            loss.backward()
            opt.step()
            sum += loss
        print(sum)
        if ench % 20 == 0:
            torch.save(model.state_dict(), "../models/" + str(ench) + ".pkl")


def test(dada_loader):
    model = MD.vgg19(pretrained=False)
    num_input = model.classifier[6].in_features
    feature_model = list(model.classifier.children())
    feature_model.pop()
    feature_model.append(nn.Linear(num_input, 7))
    model.classifier = nn.Sequential(*feature_model)
    # 加载训练过的模型进行测试
    model.load_state_dict(torch.load(""))
    model = model.cuda()
    for i, data in enumerate(dada_loader):
        img, targe = data[1]
        targe = targe.cuda()
        img = img.cuda()
        output = model(img)
        _, pred = torch.max(output.data, 1)
        print(torch.sum(pred == targe))


if __name__ == '__main__':
    trainpath="../train"
    trainimg=Make_data(trainpath)
    train(trainimg)


    testpath = "../test"
    testimg = Make_data(trainpath)
    test(testimg)