if args.net == 'vgg16':
fasterRCNN = vgg16(imdb.classes,
pretrained=True,
class_agnostic=args.class_agnostic,
lc=args.lc)
elif args.net == 'res101':
fasterRCNN = resnet(imdb.classes,
101,
pretrained=True,
class_agnostic=args.class_agnostic,
lc=args.lc)
elif args.net == 'prefood':
fasterRCNN = PreResNet50Attention(imdb.classes,
pretrained=True,
class_agnostic=args.class_agnostic,
lc=args.lc,
gc=False)
else:
print("network is not defined")
pdb.set_trace()
fasterRCNN.create_architecture()
lr = cfg.TRAIN.LEARNING_RATE
lr = args.lr
# tr_momentum = cfg.TRAIN.MOMENTUM
# tr_momentum = args.momentum
params = []
def init_network(args):
# initilize the network here.
if args.net == 'vgg16':
fasterRCNN = vgg16(imdb.classes,
pretrained=args.pretrained,
class_agnostic=args.class_agnostic)
elif args.net == 'res101':
fasterRCNN = resnet(imdb.classes,
101,
pretrained=args.pretrained,
class_agnostic=args.class_agnostic)
elif args.net == 'res50':
fasterRCNN = resnet(imdb.classes,
50,
pretrained=args.pretrained,
class_agnostic=args.class_agnostic)
elif args.net == 'res152':
fasterRCNN = resnet(imdb.classes,
152,
pretrained=args.pretrained,
class_agnostic=args.class_agnostic)
elif args.net == 'foodres50':
fasterRCNN = PreResNet50(imdb.classes,
pretrained=args.pretrained,
class_agnostic=args.class_agnostic,
weight_file=args.weight_file,
fixed_layer=args.fixed_layer)
elif args.net == 'foodres50attention':
fasterRCNN = PreResNet50Attention(imdb.classes,
pretrained=args.pretrained,
class_agnostic=args.class_agnostic,
weight_file=args.weight_file,
fixed_layer=args.fixed_layer)
elif args.net == 'foodres502fc':
fasterRCNN = PreResNet502Fc(imdb.classes,
pretrained=args.pretrained,
class_agnostic=args.class_agnostic,
weight_file=args.weight_file,
fixed_layer=args.fixed_layer)
elif args.net == 'foodres50meta':
fasterRCNN = PreResNet50Meta([0 for _ in range(170)],
pretrained=args.pretrained,
class_agnostic=args.class_agnostic,
weight_file=args.weight_file,
fixed_layer=args.fixed_layer)
elif args.net == 'foodres50_hierarchy':
main_classes = get_main_cls(imdb.classes)
fasterRCNN = PreResNet50Hierarchy(main_classes,
imdb.classes,
pretrained=args.pretrained,
class_agnostic=args.class_agnostic,
weight_file=args.weight_file,
fixed_layer=args.fixed_layer)
elif 'foodres50_hierarchy_casecade' in args.net:
nets_param = args.net.split('_')
if len(nets_param) == 3:
casecade_type = 'add_score'
alpha = 0.5
elif len(nets_param) == 5:
casecade_type = "".join(nets_param[3:5])
alpha = 0.5
elif len(nets_param) == 6:
casecade_type = "".join(nets_param[3:5])
alpha = float(nets_param[5])
main_classes = get_main_cls(imdb.classes)
fasterRCNN = PreResNet50HierarchyCasecade(
main_classes,
imdb.classes,
pretrained=args.pretrained,
class_agnostic=args.class_agnostic,
weight_file=args.weight_file,
fixed_layer=args.fixed_layer,
casecade_type=casecade_type,
alpha=alpha)
else:
print("network is not defined")
pdb.set_trace()
return fasterRCNN
pretrained=True,
class_agnostic=args.class_agnostic,
lc=args.lc,
gc=args.gc)
elif args.net == 'res50':
fasterRCNN = resnet(imdb.classes,
50,
pretrained=True,
class_agnostic=args.class_agnostic,
context=args.context)
elif args.net == 'prefood':
fasterRCNN = PreResNet50Attention(imdb.classes,
pretrained=True,
class_agnostic=args.class_agnostic,
fixed_layer=args.fixed_layer,
lc=args.lc,
gc=args.gc)
else:
print("network is not defined")
pdb.set_trace()
fasterRCNN.create_architecture()
lr = cfg.TRAIN.LEARNING_RATE
lr = args.lr
# tr_momentum = cfg.TRAIN.MOMENTUM
# tr_momentum = args.momentum
params = []