How can I save a tensorflow3D model with semantic segmentation?

Research & Models
, ,
#1

Hi, I’m new to Tensorflow and even Python. I’m interested in to use Tensorflow3D models to recognize something on mobile devices.

Before starting my problem question, can I use a tensorflow3D model on a mobile device in the first place? If so, I have a problem.

My problem is that I can’t save a model that is using semantic segmentation model, named “semantic_segmentation_model.” I’m struggled with saving the model just adding below code.

model.fit(
    x=inputs,
    callbacks=[backup_checkpoint_callback, checkpoint_callback],
    steps_per_epoch=FLAGS.num_steps_per_epoch,
    epochs=FLAGS.num_epochs,
    verbose=1 if FLAGS.run_functions_eagerly else 2)

model.close_writer()

# Added this by edom18
model.save("fileName")

This added code is in google-reserch/tf3d/train.py. I just added last code after calling fit method. Then I got some errors like below.

ValueError: batch_size is unknown at graph construction time.
  In call to configurable 'train' (<function train at 0x7f3e476aee60>)

What should I know about that?

1 Like
#2

@edom18 Welcome to Tensorflow Forum!

DO you want to take a look at this Semantic Segmentation Tutorial created by our team.
LEt us know if this resolves the issue.