def main():
# ------------------------------------------------ Training Phase ------------------------------------------------
# image_files = random.sample(glob.glob('E:\\work\\pedestrian_crop_python_process\\Pedestrain_cropDB\\train\\0\\*.bmp'), 10)
# image_files = random.sample(glob.glob('data/0.normal/*.bmp'), 10)
# data_in = data_read(image_files)
opt = Options().parse()
opt.iwidth = map_x_size
opt.iheight = map_y_size
#---new--- depth for size
ctinit = map_x_size
while ctinit > 4:
ctinit = ctinit / 2
opt.ctinit = int(ctinit)
#---new---
opt.batchsize = 64
opt.epochs = 1000
opt.mask = 0 # 1: masking for simulation map
opt.time = datetime.now()
train_dataloader = load_data(
'./data/unsupervised/train/') # path to trainset
result_path = './results/{0}/'.format(
opt.time) # reconstructions durnig the training
if not os.path.isdir(result_path):
os.mkdir(result_path)
# dataloader = load_data(opt, data_in)
model = AAE_basic(opt, train_dataloader)
model.train()
def startTest(self,params):
params['signalInfo'].emit(0, "开始检测...")
dataset = params['modelName'] # 'cus_mnist_2'
dataroot = params['path'] # 'E:\ProjectSet\Pycharm\WAIBAO\cus_mnist2'
opt = Options().parse(dataset)
opt.isTrain = False
opt.load_weights = True
opt.signal = params['signal']
opt.signalInfo = params['signalInfo']
opt.lr = self.modelsData[dataset]['opt']['lr']
opt.nz = self.modelsData[dataset]['opt']['nz']
opt.batchsize = self.modelsData[dataset]['opt']['batchsize']
opt.dataroot = dataroot
print(opt)
self.modelTest = MyTest(opt, [self.modelsData[dataset]['minVal'], self.modelsData[dataset]['maxVal'], self.modelsData[dataset]['proline']])
self.modelTest.start()
def startTrain(self, params):
params['signalInfo'].emit(0, "开始训练...")
dataset = params['name'] #'cus_mnist_2'
dataroot = params['path'] #'E:\ProjectSet\Pycharm\WAIBAO\cus_mnist2'
opt = Options().parse(dataset)
opt.signal = params['signal']
opt.load_weights = False
opt.signalInfo = params['signalInfo']
opt.lr = params['-lr']
opt.batchsize = params['-batchsize']
opt.niter = params['-niter']
opt.nz = params['-nz']
opt.desc = params['info']
opt.dataroot = dataroot
# opt.isize = 128
print(opt)
self.modelTrain = MyModel(opt)
self.modelTrain.start()
def main():
""" Training
"""
opt = Options().parse()
opt.print_freq = opt.batchsize
seed(opt.manualseed)
print("Seed:", str(torch.seed()))
if opt.phase == "inference":
opt.batchsize=1
data = load_data(opt)
model = load_model(opt, data)
if opt.phase == "inference":
model.inference()
else:
if opt.path_to_weights:
model.test()
else:
train_start = time.time()
model.train()
train_time = time.time() - train_start
print (f'Train time: {train_time} secs')
def main():
""" Training
"""
path = '/mnt/AbnormalResult/'
exps = [os.path.join(path, '1_cifar/1_pairs_airplane_2/train')]
for exp in exps:
optfile = os.path.join(exp, 'opt.txt')
opt = Options().parse_from_file(optfile)
opt.batchsize = 64
if opt.setting == 'mxn':
model = model_mxn(opt)
else:
model = model_mpairs(opt)
for iter in range(opt.niter):
weight_path = {
'net_G':
sorted(
glob.glob(
os.path.join(exp, 'weights',
'Net_G*_epoch_%d.pth*' % iter))),
'net_D':
sorted(
glob.glob(
os.path.join(exp, 'weights',
'Net_D*_epoch_%d.pth*' % iter)))
}
if len(weight_path['net_D']) != opt.n_MC_Disc and len(
weight_path['net_G']) != opt.n_MC_Gen:
continue
try:
model.load_weight(weight_path)
except:
continue
print("{}_{}".format(opt.name, iter))
model.compute_epoch(iter)
batch_size=64,
shuffle=True,
drop_last=True,
num_workers=8)
return dataloader
map_x_size = 64
map_y_size = 64
map_layer_num = 3
opt = Options().parse()
opt.iwidth = map_x_size
opt.iheight = map_y_size
opt.batchsize = 1
opt.ngpu = 0
opt.gpu_ids = -1
# ---new---
ctinit = map_x_size
while ctinit > 4:
ctinit = ctinit / 2
opt.ctinit = int(ctinit)
# ---new---
# opt.mask = 1
model_saved = False
d_loss = None
g_loss = None