run_models_on_directory passes non-standard batch_shape to model_fn about deepcell-tf HOT 4 CLOSED

The master branch now passes input_shape by default.

HOWEVER, this issue is not closed, as different model_function in the model_zoo take different parameters. We need a way to pass different model params into the model instantiation in run_models_on_directory.

from deepcell-tf.

Run_models is also generally inflexible. Will need another function or to significantly modify this function for 3d data.

from deepcell-tf.

Clarifying: this function should be able to be passed a general model function from the model_zoo. However, these models take varied parameters. If the current parameters are not what the model_function requires, it will fail.

from deepcell-tf.

Several months later, this function has not been used in months. I propose that this should be removed from running.py. This would also drastically improve the code coverage 👍

from deepcell-tf.