dlpy.model.Model.from_keras_model¶
-
classmethod
Model.
from_keras_model
(conn, keras_model, output_model_table=None, offsets=None, std=None, scale=1.0, max_num_frames=-1, include_weights=False, input_weights_file=None, verbose=False)¶ Generate a model object from a Keras model object
Parameters: - conn : CAS
The CAS connection object.
- keras_model : keras_model object
Specifies the keras model to be converted.
- output_model_table : string or dict or CAS table, optional
Specifies the CAS table to store the deep learning model.
Default: None- offsets : list, optional
Specifies the values to be subtracted from the pixel values of the input data, used if the data is an image.
- std : list or None
The pixel values of the input data are divided by these values, used if the data is an image.
- scale : float, optional
Specifies the scaling factor to apply to each image.
- max_num_frames : int, optional
Maximum number of frames for sequence processing.
- include_weights : bool, optional
Specifies whether to load the weights of the keras model.
Default: True- input_weights_file : string, optional
A fully specified client side path to the HDF5 file that stores the keras model weights. Only effective when include_weights=True. If None is given, the current weights in the keras model will be used.
Default: None- verbose : boolean optional
Specifies whether to print warning messages and debugging information
Default: False
Returns: - Model
- boolean : use GPU