otx.algorithms.common.configs#

Configs Initialization of OTX Common Algorithms.

Classes

BaseConfig(header[, description, ...])

BaseConfig Class for OTX Algorithms.

TrainType(value)

TrainType for OTX Algorithms.

LearningRateSchedule(value)

LearningRateSchedule for OTX Algorithms.

POTQuantizationPreset(value)

This Enum represents the quantization preset for post training optimization.

class otx.algorithms.common.configs.BaseConfig(header: str, description: str = 'Default parameter group description', visible_in_ui: bool = True, tiling_parameters: BaseTilingParameters = _Nothing.NOTHING, *, id: str | ID | None = ID())[source]#

Bases: ConfigurableParameters

BaseConfig Class for OTX Algorithms.

Method generated by attrs for class BaseConfig.

class BaseAlgoBackendParameters(header: str, description: str = 'Default parameter group description', visible_in_ui: bool = True, train_type: str | ConfigurableEnumTypeVar = TrainType.Incremental, mem_cache_size: int = 0, storage_cache_scheme: str | ConfigurableEnumTypeVar = StorageCacheScheme.NONE)[source]#

Bases: ParameterGroup

BaseAlgoBackendParameters for OTX Algorithms.

Method generated by attrs for class BaseConfig.BaseAlgoBackendParameters.

class BaseLearningParameters(header: str, description: str = 'Default parameter group description', visible_in_ui: bool = True, batch_size: int = 5, inference_batch_size: int = 1, unlabeled_batch_size: int = 5, num_iters: int = 1, learning_rate: float = 0.01, learning_rate_warmup_iters: int = 100, use_adaptive_interval: bool = False, enable_early_stopping: bool = True, early_stop_start: int = 3, early_stop_patience: int = 5, early_stop_iteration_patience: int = 0, num_workers: int = 0, enable_supcon: bool = False, auto_adapt_batch_size: str | ConfigurableEnumTypeVar = BatchSizeAdaptType.NONE, auto_num_workers: bool = False, input_size: str | ConfigurableEnumTypeVar = InputSizePreset.DEFAULT)[source]#

Bases: ParameterGroup

BaseLearningParameters for OTX Algorithms.

Method generated by attrs for class BaseConfig.BaseLearningParameters.

class BaseNNCFOptimization(header: str, description: str = 'Default parameter group description', visible_in_ui: bool = True, enable_quantization: bool = True, enable_pruning: bool = False, pruning_supported: bool = False, maximal_accuracy_degradation: float = 1.0)[source]#

Bases: ParameterGroup

BaseNNCFOptimization for OTX Algorithms.

Method generated by attrs for class BaseConfig.BaseNNCFOptimization.

class BasePOTParameter(header: str, description: str = 'Default parameter group description', visible_in_ui: bool = True, stat_subset_size: int = 300, stat_requests_number: int = 0, preset: str | ConfigurableEnumTypeVar = POTQuantizationPreset.PERFORMANCE)[source]#

Bases: ParameterGroup

BasePOTParameter for OTX Algorithms.

Method generated by attrs for class BaseConfig.BasePOTParameter.

class BasePostprocessing(header: str, description: str = 'Default parameter group description', visible_in_ui: bool = True, result_based_confidence_threshold: bool = True, confidence_threshold: float = 0.35, nms_iou_threshold: float = 0.5, max_num_detections: int = 0, use_ellipse_shapes: bool = False)[source]#

Bases: ParameterGroup

BasePostprocessing for OTX Algorithms.

Method generated by attrs for class BaseConfig.BasePostprocessing.

class BaseTilingParameters(description: str = 'Default parameter group description', visible_in_ui: bool = True, enable_tiling: bool = False, enable_tile_classifier: bool = False, enable_adaptive_params: bool = True, tile_size: int = 400, tile_overlap: float = 0.2, tile_max_number: int = 1500, tile_ir_scale_factor: float = 1.0, tile_sampling_ratio: float = 1.0, object_tile_ratio: float = 0.03, *, header: str = 'Tiling Parameters')[source]#

Bases: ParameterGroup

BaseTilingParameters for OTX Algorithms.

Method generated by attrs for class BaseConfig.BaseTilingParameters.

class otx.algorithms.common.configs.LearningRateSchedule(value)[source]#

Bases: ConfigurableEnum

LearningRateSchedule for OTX Algorithms.

class otx.algorithms.common.configs.POTQuantizationPreset(value)[source]#

Bases: ConfigurableEnum

This Enum represents the quantization preset for post training optimization.

class otx.algorithms.common.configs.TrainType(value)[source]#

Bases: ConfigurableEnum

TrainType for OTX Algorithms.