file_dir = 'F:/5.datasets/ip102_v2.2'
for k, v in plant.items():
train_dataset = Dataset_IP102_Augmentation(file_dir,
train=True,
transforms=data_tf,
category=v)
print('训练集类别为', k, '的图片数量有：', train_dataset.__len__())
train_data = t.utils.data.DataLoader(
train_dataset,
batch_size=16, #14
shuffle=True,
drop_last=True)
test_dataset = Dataset_IP102_Augmentation(file_dir,
train=False,
transforms=data_tf,
category=v)
test_data = t.utils.data.DataLoader(
test_dataset,
batch_size=16, #14
shuffle=True,
drop_last=True)
net = densenet(3, 102)
optimizer = t.optim.SGD(net.parameters(), lr=0.01)
criterion = t.nn.CrossEntropyLoss()
net = utils.train(net, train_data, test_data, 25, optimizer, criterion,
'densenet' + '_on_' + k)
print('\n')
x = (x - 0.5) / 0.5 #标准化
x = x.transpose((2, 0, 1)) #将channel放到第一维，只是pytorch要求的输入方式
x = torch.from_numpy(x)
return x
if __name__ == '__main__':
#输入输出形状相同
test_net = residual_block(3, 32, False)
test_x = Variable(torch.zeros(1, 3, 96, 96))
print('input:{}'.format(test_x.shape))
test_y = test_net(test_x)
print('output:{}'.format(test_y.shape))
train_set = CIFAR10('../data', train=True, transform=data_tf)
train_data = torch.utils.data.DataLoader(train_set,
batch_size=64,
shuffle=True)
test_set = CIFAR10('../data', train=False, transform=data_tf)
test_data = torch.utils.data.DataLoader(test_set,
batch_size=128,
shuffle=True)
net = resnet(3, 10)
optimizer = torch.optim.SGD(net.parameters(), lr=0.01)
criterion = nn.CrossEntropyLoss()
utils.train(net, train_data, test_data, 20, optimizer, criterion)
import os, sys
#得到当前根目录
o_path = os.getcwd()
sys.path.append(o_path)
from vgg_net import utils
from IP102.dataset_ip102 import Dataset_IP102, transform
file_dir = 'F:/5.datasets/ip102_v1.1/'
train_dataset = Dataset_IP102(file_dir, train=True, transforms=data_tf)
train_data = t.utils.data.DataLoader(
train_dataset,
batch_size=64, #14
shuffle=True,
drop_last=True)
test_dataset = Dataset_IP102(file_dir, train=False, transforms=data_tf)
test_data = t.utils.data.DataLoader(
test_dataset,
batch_size=128, #14
shuffle=True,
drop_last=True)
net = vgg_IP102()
optimizer = t.optim.SGD(net.parameters(), lr=0.01)
criterion = t.nn.CrossEntropyLoss()
utils.train(net, train_data, test_data, 20, optimizer, criterion, 'vgg_net')
from IP102.dataset_ip102_category import Dataset_IP102, plant
file_dir = 'F:/5.datasets/ip102_v1.1'
for k, v in plant.items():
train_dataset = Dataset_IP102(file_dir,
train=True,
transforms=data_tf,
category=v)
train_data = t.utils.data.DataLoader(
train_dataset,
batch_size=16, #14
shuffle=True,
drop_last=True)
test_dataset = Dataset_IP102(file_dir,
train=False,
transforms=data_tf,
category=v)
test_data = t.utils.data.DataLoader(
test_dataset,
batch_size=16, #14
shuffle=True,
drop_last=True)
net = vgg_IP102()
optimizer = t.optim.SGD(net.parameters(), lr=0.01)
criterion = t.nn.CrossEntropyLoss()
utils.train(net, train_data, test_data, 50, optimizer, criterion,
'vgg_net' + '_on_' + k)
print('\n')
drop_last=True)
net = densenet(3,102)
optimizer = t.optim.SGD(net.parameters(),lr=0.01)
criterion = t.nn.CrossEntropyLoss()
net = utils.train(net,train_data,test_data,25,optimizer,criterion,'densenet'+ '_on_'+ k)
print('\n')
'''
for k,v in plant.items():
train_dataset = Dataset_IP102_Augmentation(file_dir,train=True,transforms=data_tf,category=v)
train_data = t.utils.data.DataLoader(train_dataset,
batch_size=16,#14
shuffle=True,
drop_last=True)
test_dataset = Dataset_IP102(file_dir,train=False,transforms=data_tf,category=v)
test_data = t.utils.data.DataLoader(test_dataset,
batch_size=16,#14
shuffle=True,
drop_last=True)
net = densenet(3,102)
optimizer = t.optim.SGD(net.parameters(),lr=0.01)
criterion = t.nn.CrossEntropyLoss()
net = utils.train(net,train_data,test_data,1,optimizer,criterion,'densenet'+ '_on_augmentation_'+ k)