代码逐一学习.修改识别参数,计算最优识别
import os, sys, glob, shutil, json
os.environ["CUDA_VISIBLE_DEVICES"] ='0'
import cv2
from PILimport Image
import numpyas np
from tqdmimport tqdm, tqdm_notebook
import torch
torch.manual_seed(0)
torch.backends.cudnn.deterministic =False
torch.backends.cudnn.benchmark =True
import torchvision.modelsas models
import torchvision.transformsas transforms
import torchvision.datasetsas datasets
import torch.nnas nn
import torch.nn.functionalas F
import torch.optimas optim
from torch.autogradimport Variable
from torch.utils.data.datasetimport Dataset
class SVHNDataset(Dataset):#继承自Dataset的类
def __init__(self, img_path, img_label, transform=None):
self.img_path = img_path
self.img_label = img_label
if transformis not None:
self.transform = transform
else:
self.transform =None
def __getitem__(self, index):
img = Image.open(self.img_path[index]).convert('RGB')
if self.transformis not None:
img =self.transform(img)
lbl = np.array(self.img_label[index],dtype=np.int)
lbl =list(lbl) + (5 -len(lbl)) * [10]
return img, torch.from_numpy(np.array(lbl[:5]))
def __len__(self):
return len(self.img_path)
train_path = glob.glob('C:\\Users\\Zhyang\\Desktop\\match\\mchar_train\\*.png')#glob.glob() 函数,查找符合特定规则的文件路径名
train_path.sort()#对List进行排序.正序
train_json = json.load(open(r'C:\Users\Zhyang\Desktop\match\mchar_train.json'))
train_label = [train_json[x]['label']for xin train_json]#json中图片对应结果
# print(len(train_path), len(train_label))
train_loader = torch.utils.data.DataLoader(
SVHNDataset(train_path, train_label,
transforms.Compose([#
transforms.Resize((64,128)),#h*w图像变换 torchvision.transforms.Resize(size, interpolation=2)功能：重置图像分辨率
transforms.RandomCrop((60,120)),#随机裁剪
transforms.ColorJitter(0.3,0.3,0.2),#修改亮度、对比度和饱和度：transforms.ColorJitter
transforms.RandomRotation(45),#随机旋转
transforms.ToTensor(),#图像变换,转为tensor，并归一化至[0-1]
transforms.Normalize([0.485,0.456,0.406], [0.229,0.224,0.225])#图像变换,标准化
])),#dataset，#数据加载
batch_size=40,#batch_size（int，optional） - 每个批次要加载的样本数量（默认值:)1。
shuffle=True,#shuffle（bool，optional） - 设置为True在每个重新调整数据（默认值:) False。
num_workers=5,#num_workers（int，optional） - 用于数据加载的子进程数。0表示数据将加载到主进程中。（默认值：0）
)
val_path = glob.glob('C:\\Users\\Zhyang\\Desktop\\match\\mchar_val\\*.png')
val_path.sort()
val_json = json.load(open(r'C:\Users\Zhyang\Desktop\match\mchar_val.json'))
val_label = [val_json[x]['label']for xin val_json]
# print(len(val_path), len(val_label))
val_loader = torch.utils.data.DataLoader(
SVHNDataset(val_path, val_label,
transforms.Compose([
transforms.Resize((64,128)),
# h*w图像变换 torchvision.transforms.Resize(size, interpolation=2)功能：重置图像分辨率
transforms.RandomCrop((60,120)),# 随机裁剪
transforms.ColorJitter(0.3,0.3,0.2),# 修改亮度、对比度和饱和度：transforms.ColorJitter
transforms.RandomRotation(45),# 随机旋转
transforms.ToTensor(),# 图像变换,转为tensor，并归一化至[0-1]
transforms.Normalize([0.485,0.456,0.406], [0.229,0.224,0.225])# 图像变换,标准化
])),# dataset，#数据加载
batch_size=40,
shuffle=False,
num_workers=5,
)
class SVHN_Model1(nn.Module):
def __init__(self):
super(SVHN_Model1,self).__init__()
model_conv = models.resnet18(pretrained=True)
model_conv.avgpool = nn.AdaptiveAvgPool2d(1)
model_conv = nn.Sequential(*list(model_conv.children())[:-1])
self.cnn = model_conv
self.fc1 = nn.Linear(512,11)
self.fc2 = nn.Linear(512,11)
self.fc3 = nn.Linear(512,11)
self.fc4 = nn.Linear(512,11)
self.fc5 = nn.Linear(512,11)
def forward(self, img):
feat =self.cnn(img)
# print(feat.shape)
feat = feat.view(feat.shape[0], -1)
c1 =self.fc1(feat)
c2 =self.fc2(feat)
c3 =self.fc3(feat)
c4 =self.fc4(feat)
c5 =self.fc5(feat)
return c1, c2, c3, c4, c5
def train(train_loader, model, criterion, optimizer,epoch):#模型，准则，优化器，时代
# 切换模型为训练模式
model.train()
train_loss = []
for i, (input, target)in enumerate(train_loader):#enumerate() 函数用于将一个可遍历的数据对象(如列表、元组或字符串)组合为一个索引序列，同时列出数据和数据下标，一般用在 for 循环当中。
# print('train',i)
if use_cuda:
input = input.cuda()
target = target.cuda()
target=target.long()
c0, c1, c2, c3, c4 = model(input)
loss = criterion(c0, target[:,0]) + \
criterion(c1, target[:,1]) + \
criterion(c2, target[:,2]) + \
criterion(c3, target[:,3]) + \
criterion(c4, target[:,4])
# loss /= 6
optimizer.zero_grad()#意思是把梯度置零，也就是把loss关于weight的导数变成0.
loss.backward()#利用backward()方法进行梯度求解
optimizer.step()#用了optimizer.step()，模型才会更新
train_loss.append(loss.item())#把字典中每对key和value组成一个元组，并把这些元组放在列表中返回。
return np.mean(train_loss)
def validate(val_loader, model, criterion):
# 切换模型为预测模型
model.eval()
val_loss = []
# 不记录模型梯度信息
with torch.no_grad():
for i, (input, target)in enumerate(val_loader):
# print('validate',i)
if use_cuda:
input = input.cuda()
target = target.cuda()
target = target.long()
c0, c1, c2, c3, c4 = model(input)
loss = criterion(c0, target[:,0]) + \
criterion(c1, target[:,1]) + \
criterion(c2, target[:,2]) + \
criterion(c3, target[:,3]) + \
criterion(c4, target[:,4])
# loss /= 6
val_loss.append(loss.item())
return np.mean(val_loss)
def predict(test_loader, model, tta=10):
model.eval()
test_pred_tta =None
# TTA 次数
for _in range(tta):
test_pred = []
with torch.no_grad():
for i, (input, target)in enumerate(test_loader):
if use_cuda:
input = input.cuda()
c0, c1, c2, c3, c4 = model(input)
if use_cuda:
output = np.concatenate([
c0.data.cpu().numpy(),
c1.data.cpu().numpy(),
c2.data.cpu().numpy(),
c3.data.cpu().numpy(),
c4.data.cpu().numpy()],axis=1)
else:
output = np.concatenate([
c0.data.numpy(),
c1.data.numpy(),
c2.data.numpy(),
c3.data.numpy(),
c4.data.numpy()],axis=1)
test_pred.append(output)
test_pred = np.vstack(test_pred)
if test_pred_ttais None:
test_pred_tta = test_pred
else:
test_pred_tta += test_pred
return test_pred_tta
#生成模型
def mode_take():
best_loss=10
for epochin range(10):
train_loss = train(train_loader, model, criterion, optimizer, epoch)
val_loss = validate(val_loader, model, criterion)
val_label = [''.join(map(str, x))for xin val_loader.dataset.img_label]
print('val_label',val_label[20])
val_predict_label = predict(val_loader, model,1)
val_predict_label = np.vstack([
val_predict_label[:, :11].argmax(1),
val_predict_label[:,11:22].argmax(1),
val_predict_label[:,22:33].argmax(1),
val_predict_label[:,33:44].argmax(1),
val_predict_label[:,44:55].argmax(1),
]).T
val_label_pred = []
for xin val_predict_label:
val_label_pred.append(''.join(map(str, x[x !=10])))
print('val_label_pred',val_label_pred[:20])
val_char_acc = np.mean(np.array(val_label_pred) == np.array(val_label))
print('Epoch: {0}, Train loss: {1} \t Val loss: {2}'.format(epoch, train_loss, val_loss))
print('Val Acc', val_char_acc)
# 记录下验证集精度
if val_loss < best_loss:
best_loss = val_loss
print('Find better model in Epoch {0}, saving model.'.format(epoch))
torch.save(model.state_dict(),'C:\\Users\\Zhyang\\Desktop\\match\\model.pt')
break
#预测并生成提交文件
def produce():
test_path = glob.glob('C:\\Users\\Zhyang\\Desktop\\match\\mchar_test_a\\*.png')
test_path.sort()
test_label = [[1]] *len(test_path)
print(len(test_path),len(test_label))
test_loader = torch.utils.data.DataLoader(
SVHNDataset(test_path, test_label,
transforms.Compose([
transforms.Resize((70,140)),
transforms.RandomCrop((60,120)),
transforms.ColorJitter(0.3,0.3,0.2),
transforms.RandomRotation(5),
transforms.ToTensor(),
transforms.Normalize([0.485,0.456,0.406], [0.229,0.224,0.225])
])),
batch_size=40,
shuffle=False,
num_workers=5,
)
# 加载保存的最优模型
model.load_state_dict(torch.load('C:\\Users\\Zhyang\\Desktop\\match\\model.pt'))
test_predict_label = predict(test_loader, model,1)
print(test_predict_label[:, :11].argmax(1))
print(test_predict_label.shape)
test_label = [''.join(map(str, x))for xin test_loader.dataset.img_label]
test_predict_label = np.vstack([#vstack(tup) ，参数tup可以是元组，列表，或者numpy数组，返回结果为numpy的数组。
test_predict_label[:, :11].argmax(1),
test_predict_label[:,11:22].argmax(1),
test_predict_label[:,22:33].argmax(1),
test_predict_label[:,33:44].argmax(1),
test_predict_label[:,44:55].argmax(1),
]).T
print('test_predict_label',test_predict_label)
test_label_pred = []
for xin test_predict_label:
test_label_pred.append(''.join(map(str, x[x !=10])))
print('test_label_pred ',test_label_pred )
print(type(test_label_pred))
import pandasas pd
df_submit = pd.read_csv('C:\\Users\\Zhyang\\Desktop\\match\\mchar_sample_submit_A.csv')
df_submit['file_code'] = test_label_pred
df_submit.to_csv('C:\\Users\\Zhyang\\Desktop\\match\\mchar_sample_submit_A.csv',index=None)
#训练与验证
if __name__ =='__main__':
model = SVHN_Model1()#
criterion = nn.CrossEntropyLoss()#损失函数用
optimizer = torch.optim.Adam(model.parameters(),0.001)# torch.optim.Adam实现Adam算法,model.parameters()获取网络的参数
# params (iterable) – 待优化参数的iterable或者是定义了参数组的dict,lr (float, 可选) – 学习率（默认：1e-3）
best_loss =1000.0
# 是否使用GPU
use_cuda =True
if use_cuda:
model = model.cuda()
mode_take()
# produce()