# model_selection is an integer. # Use 0 to select a general-purpose model (operating on a 256x256 tensor). # Use 1 to select a model (operating on a 256x144 tensor) more optimized for # landscape ...
# TF Lite model represented as a FlatBuffer.