isaychris / litter-detection-tensorflow Goto Github PK

why?

Hi, could you share the model that you trained?

Dear Team FixIT and Keep America Beautiful - Great work and thanks for sharing this project! I am following steps in the notebook but ran into a parse error when loading the frozen model with the ParseFromString function which looks like it cannot decode the file into a graph.

My code is deployed within a Google Colab notebook which includes troubleshooting efforts after reading through the project documentation, tutorials and StackOverflow. Listed below is the full stack trace; any insights on how to load the model would be appreciated and thanks in advance for your help.

Full Stack Trace:

ParseError Traceback (most recent call last)
in ()
183 # text_format.Merge(proto_b, graph_def)
184 # Error: Merge/ParseFromString both throwing parse errors
--> 185 text_format.Merge(serialized_graph, od_graph_def)
186
187 # od_graph_def.ParseFromString(serialized_graph)

/usr/local/lib/python3.6/dist-packages/google/protobuf/text_format.py in Merge(text, message, allow_unknown_extension, allow_field_number, descriptor_pool)
534 allow_unknown_extension,
535 allow_field_number,
--> 536 descriptor_pool=descriptor_pool)
537
538

/usr/local/lib/python3.6/dist-packages/google/protobuf/text_format.py in MergeLines(lines, message, allow_unknown_extension, allow_field_number, descriptor_pool)
588 allow_field_number,
589 descriptor_pool=descriptor_pool)
--> 590 return parser.MergeLines(lines, message)
591
592

/usr/local/lib/python3.6/dist-packages/google/protobuf/text_format.py in MergeLines(self, lines, message)
621 """Merges a text representation of a protocol message into a message."""
622 self._allow_multiple_scalars = True
--> 623 self._ParseOrMerge(lines, message)
624 return message
625

/usr/local/lib/python3.6/dist-packages/google/protobuf/text_format.py in _ParseOrMerge(self, lines, message)
636 tokenizer = Tokenizer(lines)
637 while not tokenizer.AtEnd():
--> 638 self._MergeField(tokenizer, message)
639
640 def _MergeField(self, tokenizer, message):

/usr/local/lib/python3.6/dist-packages/google/protobuf/text_format.py in _MergeField(self, tokenizer, message)
747 merger = self._MergeMessageField
748 else:
--> 749 tokenizer.Consume(':')
750 merger = self._MergeScalarField
751

/usr/local/lib/python3.6/dist-packages/google/protobuf/text_format.py in Consume(self, token)
1099 """
1100 if not self.TryConsume(token):
-> 1101 raise self.ParseError('Expected "%s".' % token)
1102
1103 def ConsumeComment(self):

ParseError: 1:9 : Expected ":".

Hi there,

I'm looking at your documentation, and I can't find what kind of network architecture you used for this implementation. Also, I'm curious to know if you've used transfer learning on a pre-trained network or if you've retrained the entire network from scratch.
Thanks!