default='market1501std',
choices=datasets.names())
parser.add_argument('-c', '--checkpoint', type=str, required=True)
parser.add_argument('--output', type=str, required=True)
parser.add_argument('--combine', type=str, default='123')
args = parser.parse_args()
# load data set
dataset = args.dataset
cur_path = os.getcwd()
data_dir = os.path.join(cur_path, 'data', dataset)
data = datasets.create(dataset, data_dir)
query_gallery = list(set(data.query) | set(data.gallery))
# model config
config1 = Config(batch_size=128)
config2 = Config(
model_name='densenet121', height=224, width=224, batch_size=128)
config3 = Config(model_name='resnet101', batch_size=128)
if args.combine == '123':
configs = [config1, config2, config3]
elif args.combine == '12':
configs = [config1, config2]
elif args.combine == '23':
configs = [config2, config3]
else:
raise ValueError('wrong combination')
def eval(save_dir):
mAP = []
Acc = []
type=str,
default='market1501std',
choices=datasets.names())
parser.add_argument('-a', '--arch', type=str, default='resnet50')
parser.add_argument('--height', type=int, default=256)
parser.add_argument('--width', type=int, default=128)
parser.add_argument('--batch-size', type=int, default=32)
parser.add_argument('--evaluate',
dest='evaluate',
action='store_true',
help='evaluate model on validation set')
parser.add_argument('--checkpoint', default='', type=str)
args = parser.parse_args()
config = Config(model_name=args.arch,
img_translation=None,
height=args.height,
width=args.width)
# config1.height = 224
# config1.width = 224
# config1.epochs = 200
dataset = args.dataset
cur_path = os.getcwd()
logs_dir = os.path.join(cur_path, 'logs')
data_dir = os.path.join(cur_path, 'data', dataset)
data = datasets.create(dataset, data_dir)
# train_data,untrain_data = dp.split_dataset(data.trainval, 0.2)
if args.checkpoint is not None:
model = models.create(config.model_name,
configs[view])
p_y = np.argmax(p_b, axis=1)
t_y = [c for (_, c, _, _) in data.trainval]
print(np.mean(t_y == p_y))
# evaluation current model and save it
# mu.evaluate(model,data,configs[view])
save_checkpoint(
{
'state_dict': model.state_dict(),
'epoch': step + 1,
'train_data': new_train_data
},
False,
fpath=os.path.join(configs[view].logs_dir,
configs[view].model_name,
'spaco.epoch%d' % (step + 1)))
# mkdir_if_missing(logs_pth)
# torch.save(model.state_dict(), logs_pth +
# '/spaco.epoch%d' % (step + 1))
config1 = Config()
config2 = Config(model_name='densenet121', height=224, width=224)
dataset = 'market1501std'
cur_path = os.getcwd()
logs_dir = os.path.join(cur_path, 'logs')
data_dir = os.path.join(cur_path, 'data', dataset)
data = datasets.create(dataset, data_dir)
spaco([config1, config2], data, 4)
'performance': mAP
},
False,
fpath=os.path.join(
save_dir, '%s.epoch%d' % (configs[view].model_name, step + 1)))
if step + 1 == iter_steps:
features += [
mu.get_feature(net, query_gallery, data.images_dir, configs[view], device)
]
add_ratio += 1.2
# pred_y = np.argmax(sum(pred_probs), axis=1)
acc = mu.combine_evaluate(features, data)
print(acc)
config1 = Config(model_name='resnet50', loss_name='weight_softmax')
config2 = Config(model_name='densenet121',
loss_name='weight_softmax',
height=224,
width=224)
config3 = Config(model_name='resnet101',
loss_name='weight_softmax',
img_translation=2)
dataset = args.dataset
cur_path = os.getcwd()
logs_dir = os.path.join(cur_path, 'logs')
data_dir = os.path.join(cur_path, 'data', dataset)
data = datasets.create(dataset, data_dir)
spaco([config1, config2],
help="input width, default: 128 for resnet*, "
"224 for densenet*")
args = parser.parse_args()
# prepare dataset
dataset = args.dataset
cur_path = os.getcwd()
logs_dir = os.path.join(cur_path, "logs")
data_dir = os.path.join(cur_path,'data',dataset)
data = datasets.create(dataset,data_dir)
train_data = data.trainval
# model config
config = Config()
config.num_features = 512
config.width = args.width
config.height = args.height
config.set_training(False)
config.model_name = args.arch
# create model
model = models.create(config.model_name, num_features=config.num_features,
dropout=config.dropout, num_classes=config.num_classes)
model = torch.nn.DataParallel(model).cuda()
# model = model.cuda()
#epoch analysis
for i in range(0, 5):
# load model weights
mu.get_feature(net, query_gallery, data.images_dir,
configs[view], device)
]
# update training data
pred_y = np.argmax(sum(pred_probs), axis=1)
add_id = sum(add_ids)
if args.tricks:
add_ratio += 1.2
new_train_data, _ = dp.update_train_untrain(
add_id, train_data, untrain_data, pred_y)
else:
if len(untrain_data) == 0:
break
new_train_data, untrain_data = dp.update_train_untrain(
add_id, new_train_data, untrain_data, pred_y)
acc = mu.combine_evaluate(features, data)
print(acc)
config1 = Config()
config2 = Config(model_name='densenet121', height=224, width=224)
config3 = Config(model_name='resnet101', img_translation=2)
dataset = 'market1501std'
cur_path = os.getcwd()
logs_dir = os.path.join(cur_path, 'logs')
data_dir = os.path.join(cur_path, 'data', dataset)
data = datasets.create(dataset, data_dir)
cotrain([config2, config3], data, 5)
weights[view] = get_weights(pred_probs[view], pred_y, train_data, add_ratio, gamma)
sel_ids[view] = weights[view] > 0
# calculate predict probility on all data
p_b = mu.predict_prob(model, data.trainval, data_dir, configs[view])
p_y = np.argmax(p_b, axis=1)
t_y = [c for (_,c,_,_) in data.trainval]
print(np.mean(t_y == p_y))
# evaluation current model and save it
# mu.evaluate(model,data,configs[view])
save_checkpoint({
'state_dict': model.state_dict(),
'epoch': step +1,
'train_data': new_train_data}, False,
fpath = os.path.join(configs[view].logs_dir, configs[view].model_name, 'soft_spaco.epoch%d' % (step + 1))
)
# mkdir_if_missing(logs_pth)
# torch.save(model.state_dict(), logs_pth +
# '/spaco.epoch%d' % (step + 1))
config1 = Config(loss_name='weight_softmax')
config2 = Config(model_name='densenet121', loss_name='weight_softmax',
height=224, width=224)
dataset = 'market1501std'
cur_path = os.getcwd()
logs_dir = os.path.join(cur_path, 'logs')
data_dir = os.path.join(cur_path,'data',dataset)
data = datasets.create(dataset, data_dir)
soft_spaco([config1,config2], data, 5)
