Error when running the BOP challenge evaluation

[tolsicsse]

Hi
I have installed the AAE and manage to get it running, but it seems to be something wrong with the pertained BOP model since I get an error when I start he evaluation.

In the downloaded model I have the following files:

checkpoint* chkpt-800000.index*
chkpt-800000.data-00000-of-00001* chkpt-800000.meta*

But I get the following error when I run:

python ../AugmentedAutoencoder/auto_pose/m3_interface/compute_bop_results_m3.py ../AugmentedAutoencoder/auto_pose/ae/cfg_m3vision/m3_config_lmo_mp.cfg --eval_name test --dataset_name=lmo --datasets_path=. --result_folder ./result

using glfw
/root/aae_workspace/experiments/multi_object/bop_except_itodd/bop_except_itodd.cfg
(?, 128, 128, 3)
WARNING:tensorflow:From /root/.local/lib/python3.6/site-packages/auto_pose/ae/encoder.py:82: conv2d (from tensorflow.python.layers.convolutional) is deprecated and will be removed in a future version.
Instructions for updating:
Use keras.layers.conv2d instead.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.
Instructions for updating:
Colocations handled automatically by placer.
WARNING:tensorflow:From /root/.local/lib/python3.6/site-packages/auto_pose/ae/encoder.py:85: batch_normalization (from tensorflow.python.layers.normalization) is deprecated and will be removed in a future version.
Instructions for updating:
Use keras.layers.batch_normalization instead.
WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/contrib/layers/python/layers/layers.py:1624: flatten (from tensorflow.python.layers.core) is deprecated and will be removed in a future version.
Instructions for updating:
Use keras.layers.flatten instead.
WARNING:tensorflow:From /root/.local/lib/python3.6/site-packages/auto_pose/ae/encoder.py:99: dense (from tensorflow.python.layers.core) is deprecated and will be removed in a future version.
Instructions for updating:
Use keras.layers.dense instead.

NotFoundError Traceback (most recent call last)
~/AugmentedAutoencoder/auto_pose/m3_interface/compute_bop_results_m3.py in
91 if pose_est_method:
92 # mp_pose_estimator = AePoseEstimator(os.path.join(workspace_path,'cfg_m3vision/test_config.cfg'))
---> 93 mp_pose_estimator = MPPoseEstimator(m3_config_path)
94
95 if pose_refiner_method:

~/.local/lib/python3.6/site-packages/auto_pose/m3_interface/mp_pose_estimator.py in init(self, test_config_path)
62 checkpoint_file = utils.get_checkpoint_basefilename(log_dir, False, latest=self.train_args.getint('Training', 'NUM_ITER'), joint=True)
63
---> 64 self.codebook_multi, self.dataset = ae_factory.build_codebook_from_name(experiment_name, experiment_group, return_dataset=True, joint=True)
65 encoder = self.codebook_multi._encoder
66

~/.local/lib/python3.6/site-packages/auto_pose/ae/ae_factory.py in build_codebook_from_name(experiment_name, experiment_group, return_dataset, return_decoder, joint)
205 encoder = build_encoder(x, args)
206 if joint:
--> 207 codebook = build_codebook_multi(encoder, dataset, args, checkpoint_file)
208 else:
209 codebook = build_codebook(encoder, dataset, args)

~/.local/lib/python3.6/site-packages/auto_pose/ae/ae_factory.py in build_codebook_multi(encoder, dataset, args, checkpoint_file_basename)
161 existing_embs = []
162 if checkpoint_file_basename is not None:
--> 163 var_list = checkpoint_utils.list_variables(checkpoint_file_basename)
164 for v in var_list:
165 if 'embedding_normalized_' in v[0]:

/usr/local/lib/python3.6/dist-packages/tensorflow/contrib/framework/python/framework/checkpoint_utils.py in list_variables(checkpoint_dir)
90 List of tuples (name, shape).
91 """
---> 92 reader = load_checkpoint(checkpoint_dir)
93 variable_map = reader.get_variable_to_shape_map()
94 names = sorted(variable_map.keys())

/usr/local/lib/python3.6/dist-packages/tensorflow/contrib/framework/python/framework/checkpoint_utils.py in load_checkpoint(filepattern)
61 raise ValueError("Couldn't find 'checkpoint' file or checkpoints in "
62 "given directory %s" % filepattern)
---> 63 return train.NewCheckpointReader(filename)
64
65

/usr/local/lib/python3.6/dist-packages/tensorflow/python/pywrap_tensorflow_internal.py in NewCheckpointReader(filepattern)
324 with errors.raise_exception_on_not_ok_status() as status:
325 from tensorflow.python.util import compat
--> 326 return CheckpointReader(compat.as_bytes(filepattern), status)
327
328 NewCheckpointReader._tf_api_names_v1 = ['train.NewCheckpointReader']

/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/errors_impl.py in exit(self, type_arg, value_arg, traceback_arg)
526 None, None,
527 compat.as_text(c_api.TF_Message(self.status.status)),
--> 528 c_api.TF_GetCode(self.status.status))
529 # Delete the underlying status object from memory otherwise it stays alive
530 # as there is a reference to status from this from the traceback due to

NotFoundError: Unsuccessful TensorSliceReader constructor: Failed to find any matching files for /root/aae_workspace/experiments/multi_object/bop_except_itodd/checkpoints/chkpt-joint-800000

[tolsicsse]

I noticed that the -joint was added because an argument joint=True was set, by changing it to False it manage to go further but then it another error was raised that bop_except_itodd/embed_obj_bbs_var was missing in the checkpoint.

[MartinSmeyer]

Hey @tolsicsse,

I accidentally uploaded the model in an old format without the joint option when I last updated the download link. Now, I have corrected the download link to point to the correct model again: http://fex.dlr.de/fop/Te7peqPc/bop.zip

Please set the joint option back to False and try to run this model.

To run this model you will need >25GB RAM.

[tolsicsse]

Yes it works now. Thanks.