def __init__(self, configer):
self.configer = configer
self.runner_state = dict(iters=0,
last_iters=0,
epoch=0,
last_epoch=0,
performance=0,
val_loss=0,
max_performance=0,
min_val_loss=0)
self.batch_time = AverageMeter()
self.data_time = AverageMeter()
self.train_losses = DictAverageMeter()
self.val_losses = DictAverageMeter()
self.cls_model_manager = ModelManager(configer)
self.cls_data_loader = DataLoader(configer)
self.running_score = RunningScore(configer)
self.cls_net = self.cls_model_manager.get_model()
self.solver_dict = self.configer.get(
self.configer.get('train', 'solver'))
self.optimizer, self.scheduler = Trainer.init(self._get_parameters(),
self.solver_dict)
self.cls_net = RunnerHelper.load_net(self, self.cls_net)
self.cls_net, self.optimizer = RunnerHelper.to_dtype(
self, self.cls_net, self.optimizer)
self.train_loader = self.cls_data_loader.get_trainloader()
self.val_loader = self.cls_data_loader.get_valloader()
self.loss = self.cls_model_manager.get_loss()
def __init__(self, configer):
self.configer = configer
self.batch_time = AverageMeter()
self.data_time = AverageMeter()
self.seg_visualizer = SegVisualizer(configer)
self.loss_manager = LossManager(configer)
self.module_runner = ModuleRunner(configer)
self.model_manager = ModelManager(configer)
self.optim_scheduler = OptimScheduler(configer)
self.seg_data_loader = DataLoader(configer)
self.save_dir = self.configer.get('test', 'out_dir')
self.seg_net = None
self.test_loader = None
self.test_size = None
self.infer_time = 0
self.infer_cnt = 0
self._init_model()
def __init__(self, configer):
self.configer = configer
self.batch_time = AverageMeter()
self.foward_time = AverageMeter()
self.backward_time = AverageMeter()
self.loss_time = AverageMeter()
self.data_time = AverageMeter()
self.train_losses = AverageMeter()
self.val_losses = AverageMeter()
self.seg_visualizer = SegVisualizer(configer)
self.loss_manager = LossManager(configer)
self.module_runner = ModuleRunner(configer)
self.model_manager = ModelManager(configer)
self.data_loader = DataLoader(configer)
self.optim_scheduler = OptimScheduler(configer)
self.data_helper = DataHelper(configer, self)
self.evaluator = get_evaluator(configer, self)
self.seg_net = None
self.train_loader = None
self.val_loader = None
self.optimizer = None
self.scheduler = None
self.running_score = None
self._init_model()
def __init__(self, configer):
self.configer = configer
self.runner_state = dict(iters=0,
last_iters=0,
epoch=0,
last_epoch=0,
performance=0,
val_loss=0,
max_performance=0,
min_val_loss=0)
self.batch_time = AverageMeter()
self.data_time = AverageMeter()
self.train_losses = DictAverageMeter()
self.val_losses = DictAverageMeter()
self.cls_model_manager = ModelManager(configer)
self.cls_data_loader = DataLoader(configer)
self.running_score = RunningScore(configer)
self.cls_net = self.cls_model_manager.get_model()
self.solver_dict = self.configer.get(
self.configer.get('train', 'solver'))
self.optimizer, self.scheduler = Trainer.init(self._get_parameters(),
self.solver_dict)
self.cls_net = RunnerHelper.load_net(self, self.cls_net)
self.cls_net, self.optimizer = RunnerHelper.to_dtype(
self, self.cls_net, self.optimizer)
self.train_loaders = dict()
self.val_loaders = dict()
for source in range(self.configer.get('data', 'num_data_sources')):
self.train_loaders[source] = self.cls_data_loader.get_trainloader(
source=source)
self.val_loaders[source] = self.cls_data_loader.get_valloader(
source=source)
if self.configer.get('data', 'mixup'):
assert (self.configer.get('data', 'num_data_sources') == 2
), "mixup only support src0 and src1 load the same dataset"
self.loss = self.cls_model_manager.get_loss()
def __init__(self, configer):
self.crop_size = configer.get('train',
'data_transformer')['input_size']
val_trans_seq = [
x for x in configer.get('val_trans', 'trans_seq')
if 'random' not in x
]
configer.update(('val_trans', 'trans_seq'), val_trans_seq)
configer.get('val', 'data_transformer')['input_size'] = configer.get(
'test', 'data_transformer').get('input_size', None)
configer.update(('train', 'data_transformer'),
configer.get('val', 'data_transformer'))
configer.update(('val', 'batch_size'),
int(os.environ.get('batch_size', 16)))
configer.update(('test', 'batch_size'),
int(os.environ.get('batch_size', 16)))
self.save_dir = configer.get('test', 'out_dir')
self.dataset_name = configer.get('test', 'eval_set')
self.sscrop = configer.get('test', 'sscrop')
self.configer = configer
self.batch_time = AverageMeter()
self.data_time = AverageMeter()
self.loss_manager = LossManager(configer)
self.module_runner = ModuleRunner(configer)
self.model_manager = ModelManager(configer)
self.seg_data_loader = DataLoader(configer)
self.seg_net = None
self.test_loader = None
self.test_size = None
self.infer_time = 0
self.infer_cnt = 0
self._init_model()
pprint.pprint(configer.params_root)