pyanomaly.datatools package

Subpackages

• pyanomaly.datatools.abstract package
  • Submodules
  • pyanomaly.datatools.abstract.abstract_datasets_builder module
  • pyanomaly.datatools.abstract.abstract_datasets_factory module
  • pyanomaly.datatools.abstract.abstract_evaluate_method module
  • pyanomaly.datatools.abstract.image_dataset module
  • pyanomaly.datatools.abstract.readers module
  • pyanomaly.datatools.abstract.video_dataset module
  • Module contents
• pyanomaly.datatools.dataclass package
  • Subpackages
    • pyanomaly.datatools.dataclass.augment package
      • Submodules
      • pyanomaly.datatools.dataclass.augment.augment_api module
      • Module contents
    • pyanomaly.datatools.dataclass.sampler package
      • Submodules
      • pyanomaly.datatools.dataclass.sampler.common module
      • pyanomaly.datatools.dataclass.sampler.dist_inf_sampler module
      • pyanomaly.datatools.dataclass.sampler.inf_sampler module
      • Module contents
  • Submodules
  • pyanomaly.datatools.dataclass.avenue_ped_shanghai module
  • pyanomaly.datatools.dataclass.datasets_factory module
  • Module contents
• pyanomaly.datatools.evaluate package
  • Submodules
  • pyanomaly.datatools.evaluate.eval_function module
  • pyanomaly.datatools.evaluate.utils module
  • Module contents

Submodules

pyanomaly.datatools.datatools_api module

@author: Yuhao Cheng @contact: yuhao.cheng[at]outlook.com

class pyanomaly.datatools.datatools_api.DataAPI(cfg, is_training)

Bases: pyanomaly.datatools.abstract.abstract_datasets_builder.AbstractBuilder

The API class is used for getting the torch.data.Dataloader dictionary

NAME = 'DatasetAPI'

build()

The building method Args:

None

Returns:

dataloader_dict(OrderedDict): The dictionary contains the dataloader used in training or val/test process For example: {

'train':{

'general_data_dict':{

'all': torch.data.Dataloader

}, 'w_dataset_dict': {

'01': torch.data.Dataloader, ... ...

}, ... ...

}, 'val':{

'general_dataset_dict':{

'01': torch.data.Dataloader, '02': torch.data.Dataloader, '03': torch.data.Dataloder, ... ...

}, ... ...

}

}

class pyanomaly.datatools.datatools_api.EvaluateAPI(cfg, is_training)

Bases: object

The class to get the Evaluation object

pyanomaly.datatools.datatools_registry module

@author: Yuhao Cheng @contact: yuhao.cheng[at]outlook.com

pyanomaly.datatools.tools module

@author: Yuhao Cheng @contact: yuhao.cheng[at]outlook.com

class pyanomaly.datatools.tools.RecordResult(fpr=None, tpr=None, thresholds=None, auc=- inf, dataset=None, loss_file=None, sigma=0)

Bases: object

get_threshold()

pyanomaly.datatools.tools.collect_fn(batch)

image_b, image_a, image, image_f, label, detection_result = batch

pyanomaly.datatools.tools.collect_fn_local(batch)

image_b, image_a, image, image_f, crop_objects = batch

pyanomaly.datatools.tools.decide_back_front(dataset_path, verbose='testing_vad', duration=3)

decide the back and front, save in json Args:

dataset_path: e.g. './data/shanghaitech/training/frames' duration: step, e.g. current-3, current+3

pyanomaly.datatools.tools.make_global_db(data_path, split='testing')

Make the database based on the detections Args:

data_path: e.g. 'data/shanghaitech/normal'

pyanomaly.datatools.tools.make_objects_box_db(data_path, split='training', det_threshold=0.95, time_file='./training_3.json', verbose='none')

Make the database based on the detections Args:

data_path: e.g. 'data/shanghaitech/normal' det_threshold: 0.5

pyanomaly.datatools.tools.make_objects_db(data_path, split='training', det_threshold=0.95, time_file='./training_3.json', verbose='none')

Make the database based on the detections Args:

data_path: e.g. 'data/shanghaitech/normal' det_threshold: 0.5

Module contents