help='number of workers for dataloader')
parser.add_argument('-b',
type=int,
default=16,
help='batch size for dataloader')
parser.add_argument('-s',
type=bool,
default=True,
help='whether shuffle the dataset')
args_dict = vars(parser.parse_args())
logger.info(args_dict)
net_type = args_dict['net']
use_gpu = args_dict['gpu']
net = build_network(archi=net_type, use_gpu=use_gpu)
# logger.info(net)
net.load_state_dict(torch.load(args_dict['weights']), args_dict['gpu'])
net.eval()
example_image_dir = '../datasets/Image_Search_Dataset/train'
dataset = datasets.ImageFolder(example_image_dir, transform=None)
idx_to_class = {v: k for k, v in dataset.class_to_idx.items()}
original_path = "../datasets/Image_Search_Dataset/train"
list_classes = next(os.walk(original_path))[1]
if os.path.isfile('lsh.p'):
logger.info("load indexed dict")
lsh = pickle.load(open('lsh.p', 'rb'))
feature_dict = pickle.load(open('feature_dict.p', 'rb'))
parser.add_argument('-s',
type=bool,
default=True,
help='whether shuffle the dataset')
args = parser.parse_args()
args_dict = vars(parser.parse_args())
logger.info(args_dict)
### Initialize the model
net_type = args_dict['net']
use_gpu = args_dict['gpu']
standard_folder = global_settings.TRAIN_FOLDER
list_author = next(os.walk(standard_folder))[1]
num_classes = len(list_author)
net = build_network(archi=net_type,
use_gpu=use_gpu,
num_classes=num_classes)
net.cuda()
# net_type = args_dict['net']
# use_gpu = args_dict['gpu']
# standard_folder = global_settings.TRAIN_FOLDER
# list_author = next(os.walk(standard_folder))[1]
# num_class = len(list_author)
# net = build_network(archi = net_type, use_gpu=use_gpu) # , num_class=num_class
logger.info(net)
net.load_state_dict(torch.load(args.weights), args.gpu)
net.eval()
test_image_dir = global_settings.TEST_FOLDER
### Datasets loader
stamp_training_loader = get_training_dataloader(global_settings.TRAIN_MEAN,
global_settings.TRAIN_STD,
num_workers=args.w,
batch_size=args.b,
shuffle=args.s)
stamp_test_loader, idx_to_class = get_test_dataloader(
global_settings.TRAIN_MEAN,
global_settings.TRAIN_STD,
num_workers=args.w,
batch_size=args.b,
shuffle=args.s)
### Initialize the model
net = build_network(args)
# net_type = args_dict['net']
# use_gpu = args_dict['gpu']
# standard_folder = global_settings.TRAIN_FOLDER
# list_author = next(os.walk(standard_folder))[1]
# num_classes = len(list_author)
# net = build_network(archi=net_type, use_gpu=use_gpu, num_classes=num_classes)
### Optimizer & compute loss function
loss_function = nn.CrossEntropyLoss()
optimizer = optim.SGD(net.parameters(),
lr=args.lr,
momentum=0.9,
weight_decay=5e-4)
# Log infor modularity to Pytorch models & optimizer