Dear Sir,
I've installed the SNPE 1.25.1 and there is no issue when I run "setup_inceptionv3.py" with the default model named as "inception_v3_2016_08_28_frozen.pb". And I tried to retrain as new model named as "rgb_graph.pb".
Then I run "setup_inceptionv3.py" with modifying the pb, label, and checksum, and it shows the following errors: 2019-05-10 10:04:05.850162: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x338a1d0 executing computations on platform Host. Devices: 2019-05-10 10:04:05.850181: I tensorflow/compiler/xla/service/service.cc:158] StreamExecutor device (0): , 2019-05-10 10:04:06,202 - 106 - ERROR - Conversion failed: ERROR_TF_NODE_NOT_FOUND_IN_GRAPH: Node not found in graph. Node name: input
Then I modified the function convert_to_dlc in "setup_inceptionv3.py" and run it:
'--input_dim', 'input', '1,299,299,3', ==> '--input_dim', 'Placeholder', '1,299,299,3',
It shows: 2019-05-10 09:51:39.609519: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x3d77b80 executing computations on platform Host. Devices: 2019-05-10 09:51:39.609537: I tensorflow/compiler/xla/service/service.cc:158] StreamExecutor device (0): , 2019-05-10 09:51:40,025 - 109 - ERROR - Encountered Error: 'InceptionV3/Predictions/Reshape_1' Traceback (most recent call last): File "/home/han/project/SNPE/snpe-1.25.1.310/bin/x86_64-linux-clang/snpe-tensorflow-to-dlc", line 102, in main converter = DlcConverter(model, not args.allow_unconsumed_nodes) File "/home/han/project/SNPE/snpe-1.25.1.310/lib/python/snpe/converters/tensorflow/converter.py", line 289, in __init__ self._ops = self._resolve_graph_operations_from_model(model) File "/home/han/project/SNPE/snpe-1.25.1.310/lib/python/snpe/converters/tensorflow/converter.py", line 541, in _resolve_graph_operations_from_model queue = [operations_map[output_op_name]] KeyError: 'InceptionV3/Predictions/Reshape_1'
Please advise how to convert the new model (retrain) and convert to dlc file successfully. Thanks, Han