Checkpoints trained in Python2 are not compatible to load in Python3

[tryolabs]

Checkpoints trained in Python2 are not compatible to load in Python3

[arun-kumark]

Hi,
While training, I am getting the following error:
tensorflow.python.framework.errors_impl.InvalidArgumentError: Name: , Context feature 'depth' is required but could not be found.
Please help in resolving the issue.

Logs snippet is below:

  File "/home/arun/anaconda3/lib/python3.6/site-packages/tensorflow/python/framework/errors_impl.py", line 473, in __exit__
    c_api.TF_GetCode(self.status.status))
tensorflow.python.framework.errors_impl.InvalidArgumentError: Name: , Context feature 'depth' is required but could not be found.
	 [[Node: object_detection_dataset/ParseSingleSequenceExample/ParseSingleSequenceExample = ParseSingleSequenceExample[Ncontext_dense=5, Ncontext_sparse=0, Nfeature_list_dense=0, Nfeature_list_sparse=5, Tcontext_dense=[DT_INT64, DT_STRING, DT_INT64, DT_STRING, DT_INT64], context_dense_shapes=[[], [], [], [], []], context_sparse_types=[], feature_list_dense_shapes=[], feature_list_dense_types=[], feature_list_sparse_types=[DT_INT64, DT_INT64, DT_INT64, DT_INT64, DT_INT64], _device="/job:localhost/replica:0/task:0/device:CPU:0"](object_detection_dataset/ReaderReadV2:1, object_detection_dataset/ParseSingleSequenceExample/ParseSingleSequenceExample/feature_list_dense_missing_assumed_empty, object_detection_dataset/ParseSingleSequenceExample/ParseSingleSequenceExample/context_dense_keys_0, object_detection_dataset/ParseSingleSequenceExample/ParseSingleSequenceExample/context_dense_keys_1, object_detection_dataset/ParseSingleSequenceExample/ParseSingleSequenceExample/context_dense_keys_2, object_detection_dataset/ParseSingleSequenceExample/ParseSingleSequenceExample/context_dense_keys_3, object_detection_dataset/ParseSingleSequenceExample/ParseSingleSequenceExample/context_dense_keys_4, object_detection_dataset/ParseSingleSequenceExample/ParseSingleSequenceExample/feature_list_sparse_keys_0, object_detection_dataset/ParseSingleSequenceExample/ParseSingleSequenceExample/feature_list_sparse_keys_1, object_detection_dataset/ParseSingleSequenceExample/ParseSingleSequenceExample/feature_list_sparse_keys_2, object_detection_dataset/ParseSingleSequenceExample/ParseSingleSequenceExample/feature_list_sparse_keys_3, object_detection_dataset/ParseSingleSequenceExample/ParseSingleSequenceExample/feature_list_sparse_keys_4, object_detection_dataset/ParseSingleSequenceExample/Const, object_detection_dataset/ParseSingleSequenceExample/ParseSingleSequenceExample/feature_list_dense_missing_assumed_empty, object_detection_dataset/ParseSingleSequenceExample/Const, object_detection_dataset/ParseSingleSequenceExample/ParseSingleSequenceExample/feature_list_dense_missing_assumed_empty, object_detection_dataset/ParseSingleSequenceExample/Const, object_detection_dataset/ParseSingleSequenceExample/ParseSingleSequenceExample/debug_name)]]

Thanks
Kind Regards
arun

[arun-kumark]

I created the CSV file, and then created the tfrecords file:

The CSV file snap is below:

filename,width,height,class,xmin,ymin,xmax,ymax
2018-02-03_162606_477161855_3.jpg,1280,720,barilla,640,2,1280,511
2018-02-03_162606_477161855_4.jpg,1280,720,barilla,12,456,774,720
2018-02-03_162606_477161855_0.jpg,1280,720,barilla,155,82,949,720
2018-02-03_162606_477161855_2.jpg,1280,720,barilla,21,200,823,720
2018-02-03_162606_477161855_1.jpg,1280,720,barilla,648,1,1280,720

[chricke]

have the same issue - any update on this?

[chricke]

I changed my dataset so it can be transformed using the cvs_reader - that worked for me. Before that i did the conversion manually and the "depth" parameter was missing which is set in the cvs_reader on line 133 to the fixed value of 3.

I'm also getting the same error.. can someone provide detailed solution?

[dekked]

Please re-report this if it happens with the latest version on master. Closing for now.