# 验证集不增强
val_loader = torch.utils.data.DataLoader(datasets.ImageFolder(
dataPath + '/val', transform=data_transforms),
batch_size=batch_size,
shuffle=False,
num_workers=0,
pin_memory=use_gpu)
# 神经网络和优化器
from model import Net
model = Net()
model_dict = model.load_state_dict(torch.load('./model/40.pth'))
fc2_infeature = model.fc2.in_features
model.fc2 = nn.Linear(fc2_infeature, 4)
if use_gpu:
model.cuda()
# 优化器和学习率变化策略设置
optimizer = optim.Adam(filter(lambda p: p.requires_grad, model.parameters()),
lr=lr)
scheduler = optim.lr_scheduler.ReduceLROnPlateau(optimizer,
'min',
patience=5,
factor=0.5,
verbose=True)
def train(epoch):
from torch.autograd import Variable
import torch.nn.functional as F
import torchvision.datasets as datasets
import numpy as np
import time
parser = argparse.ArgumentParser(description='evaluation script')
parser.add_argument('--c', type=int, default=60,metavar='D',
help="folder where data is located. train_data.zip and test_data.zip need to be found in the folder")
args = parser.parse_args()
print('- - - - - - step3 Class inference Begin - - - - - - ')
start = time.time()
model_class = Net()
fc2_features = model_class.fc2.in_features
model_class.fc2 = nn.Linear(fc2_features, 121)
model_class.load_state_dict(torch.load('./new_model/class_121.pth'))
model_class.cuda()
model_class.eval()
model_temporary = Net()
fc2_features = model_temporary.fc2.in_features
model_temporary.fc2 = nn.Linear(fc2_features, 2)
model_temporary.load_state_dict(torch.load('./new_model/temporary_2.pth'))
model_temporary.cuda()
model_temporary.eval()
model_data = Net()
fc2_features = model_data.fc2.in_features
model_data.fc2 = nn.Linear(fc2_features, 20)
model_data.load_state_dict(torch.load('./new_model/data_20.pth'))
model_data.cuda()
model_data.eval()
edgecolors='none')
else:
ax.scatter(location[i, 0],
location[i, 1],
c=colors[l],
s=10,
alpha=0.7,
edgecolors='none')
test_loader = torch.utils.data.DataLoader(datasets.MNIST(
'../data',
train=False,
transform=transforms.Compose(
[transforms.ToTensor(),
transforms.Normalize((0.1307, ), (0.3081, ))])),
batch_size=100,
shuffle=False)
model = Net()
model = load_network(model)
model.fc2 = nn.Sequential()
model = model.eval()
model = model.cuda()
fig, ax = plt.subplots()
test(model, test_loader)
ax.grid(True)
ax.legend(loc='best')
fig.savefig('train.jpg')
# from tqdm import tqdm
import os
import PIL.Image as Image
import torch
from torch.autograd import Variable
import torch.nn.functional as F
import torchvision.datasets as datasets
import numpy as np
import time
print('- - - - - - step3 Class inference Begin - - - - - - ')
start = time.time()
model_class = Net()
fc2_features = model_class.fc2.in_features
model_class.fc2 = nn.Linear(fc2_features, 121)
model_class.load_state_dict(torch.load('./new model/class_121.pth'))
model_data = Net()
fc2_features = model_data.fc2.in_features
model_data.fc2 = nn.Linear(fc2_features, 20)
model_data.load_state_dict(torch.load('./new model/data_20.pth'))
def pil_loader(path):
# open path as file to avoid ResourceWarning (https://github.com/python-pillow/Pillow/issues/835)
with open(path, 'rb') as f:
with Image.open(f) as img:
return img.convert('RGB')
num_workers=0,
pin_memory=use_gpu)
val_loader = torch.utils.data.DataLoader(datasets.ImageFolder(
dataPath + '/val', transform=data_transforms),
batch_size=batch_size,
shuffle=False,
num_workers=0,
pin_memory=use_gpu)
# Neural Network and Optimizer
from model import Net
model = Net()
model.load_state_dict(torch.load('./model/domestic40.pth'))
fc2_features = model.fc2.in_features
model.fc2 = nn.Linear(fc2_features, 20)
if use_gpu:
model.cuda()
optimizer = optim.Adam(filter(lambda p: p.requires_grad, model.parameters()),
lr=lr)
scheduler = optim.lr_scheduler.ReduceLROnPlateau(optimizer,
'min',
patience=5,
factor=0.5,
verbose=True)
def train(epoch):
model.train()