torch.manual_seed(opt.seed)
if cuda:
torch.cuda.manual_seed(opt.seed)
print('===> Loading dataset(s)')
if opt.mode.lower() == 'train':
whole_train_set = dataset.get_whole_training_set()
whole_training_data_loader = DataLoader(dataset=whole_train_set,
num_workers=opt.threads,
batch_size=opt.cacheBatchSize,
shuffle=False,
pin_memory=cuda)
train_set = dataset.get_training_query_set(opt.margin,
opt.random_pos_level,
opt.random_crop)
print('====> Training query set:', len(train_set))
whole_test_set = dataset.get_whole_val_set()
print('===> Evaluating on val set, query count:',
whole_test_set.dbStruct.numQ)
elif opt.mode.lower() == 'test':
if opt.split.lower() == 'test':
whole_test_set = dataset.get_whole_test_set()
print('===> Evaluating on beelbank100 set')
elif opt.split.lower() == 'test250k':
whole_test_set = dataset.get_250k_test_set()
print('===> Evaluating on test250k set')
elif opt.split.lower() == 'train':
whole_test_set = dataset.get_whole_training_set()
random.seed(opt.seed)
np.random.seed(opt.seed)
torch.manual_seed(opt.seed)
if cuda:
torch.cuda.manual_seed(opt.seed)
print('===> Loading dataset(s)')
if opt.mode.lower() == 'train':
whole_train_set = dataset.get_whole_training_set()
whole_training_data_loader = DataLoader(dataset=whole_train_set,
num_workers=opt.threads,
batch_size=opt.cacheBatchSize,
shuffle=False,
pin_memory=cuda)
train_set = dataset.get_training_query_set(opt.margin)
print('====> Training query set:', len(train_set))
whole_test_set = dataset.get_whole_val_set()
print('===> Evaluating on val set, query count:',
whole_test_set.dbStruct.numQ)
elif opt.mode.lower() == 'test':
if opt.split.lower() == 'test':
whole_test_set = dataset.get_whole_test_set()
print('===> Evaluating on test set')
elif opt.split.lower() == 'test250k':
whole_test_set = dataset.get_250k_test_set()
print('===> Evaluating on test250k set')
elif opt.split.lower() == 'train':
whole_test_set = dataset.get_whole_training_set()
print('===> Evaluating on train set')
torch.cuda.manual_seed(opt.seed)
print('===> Loading dataset(s)')
if opt.mode.lower() == 'train':
# a list of train sets
whole_train_set = dataset.get_whole_training_set(opt.arch.lower())
whole_training_data_loader = [
DataLoader(dataset=set,
num_workers=opt.threads,
batch_size=opt.cacheBatchSize,
shuffle=False,
pin_memory=cuda) for set in whole_train_set
]
# a list of train sets
train_set_list = dataset.get_training_query_set(
opt.arch.lower(), opt.margin)
print('====> Training query set:', len(train_set_list[0]))
whole_test_set = dataset.get_whole_val_set(opt.arch.lower())
print('===> Evaluating on val set, query count:',
whole_test_set.dbStruct.numQ)
elif opt.mode.lower() == 'test':
if opt.split.lower() == 'test':
whole_test_set = dataset.get_whole_test_set(opt.arch.lower())
print('===> Evaluating on test set')
elif opt.split.lower() == 'test_hard':
whole_test_set = dataset.get_test_hard_set(opt.arch.lower())
print('===> Evaluating on test hard')
elif opt.split.lower() == 'test250k':
if opt.dataset.lower() == 'pittsburgh':
whole_test_set = dataset.get_250k_test_set(opt.arch.lower())