hi, I got snpe 1.15.0 : ubuntu14.04 and tensorflow 1.5
I tried the instructions of doc which are under the Examples Tutorials -> Model Conversion -> MobilenetSSD
I got the output (which don't seem to be correct):
vecen@ubuntu:~/snpe-1.15.0/models/mobilenet$ snpe-tensorflow-to-dlc --graph ssd_mobilenet_v1_coco_2017_11_17/frozen_inference_graph.pb -i Preprocessor/sub 300,300,3 --out_node detection_classes --out_node detection_boxes --out_node detection_scores --dlc mobilenet_ssd.dlc --allow_unconsumed_nodes
2018-07-04 00:57:25.550113: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.1 instructions, but these are available on your machine and could speed up CPU computations.
2018-07-04 00:57:25.550150: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations.
2018-07-04 00:57:25.550178: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.
2018-07-04 00:57:25.550186: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX2 instructions, but these are available on your machine and could speed up CPU computations.
2018-07-04 00:57:25.550190: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use FMA instructions, but these are available on your machine and could speed up CPU computations.
2018-07-04 00:57:29.062221: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[604] has already been set.
2018-07-04 00:57:29.062259: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[6536] has already been set.
2018-07-04 00:57:29.062295: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[6532] has already been set.
2018-07-04 00:57:29.062305: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[561] has already been set.
2018-07-04 00:57:29.063386: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[6535] has already been set.
2018-07-04 00:57:29.062327: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[6538] has already been set.
2018-07-04 00:57:29.065622: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[6539] has already been set.
2018-07-04 00:57:29.066364: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[6541] has already been set.
2018-07-04 00:57:29.067252: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[6543] has already been set.
2018-07-04 00:57:29.067711: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[6545] has already been set.
2018-07-04 00:57:29.067972: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[6533] has already been set.
2018-07-04 00:57:29.068300: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[565] has already been set.
2018-07-04 00:57:29.068624: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[564] has already been set.
2018-07-04 00:57:29.068884: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[562] has already been set.
2018-07-04 00:57:29.069109: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[563] has already been set.
2018-07-04 00:57:29.091033: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[569] has already been set.
2018-07-04 00:57:29.091036: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[566] has already been set.
2018-07-04 00:57:29.091036: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[568] has already been set.
2018-07-04 00:57:29.091110: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[567] has already been set.
2018-07-04 00:57:29.091126: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[571] has already been set.
2018-07-04 00:57:29.091058: W tensorflow/core/framework/op_kernel.cc:1192] Internal: Retval[570] has already been set.
2018-07-04 00:57:29,099 - 391 - WARNING - ERROR_TF_FALLBACK_TO_ONDEMAND_EVALUATION: Unable to resolve operation output shapes in single pass. Using on-demand evaluation!
anyone else got the similar output when converting to dlc ???
using: snpe-tensorflow-to-dlc --graph ssd_mobilenet_v1_coco_2017_11_17/frozen_inference_graph.pb -i Preprocessor/sub 300,300,3 --out_node detection_classes --out_node detection_boxes --out_node detection_scores --dlc mobilenet_ssd.dlc --allow_unconsumed_nodes
Hi guoguoxin2004,
Did you try to run your DLC on mobile ? When i try to read the output map of the network, only one ouput node can be resolved.
I use same command to convert mobilev1-ssd to DLC. :(
Hi svenzhang,
SNPE generates the last layer's output by default. If you want to extract other layers' output, you have to configure output layers.
If you're using C++ API, you need to call setOutputLayers() API.
https://developer.qualcomm.com/docs/snpe/group__c__plus__plus__apis.html...
If you're using snpe-net-run tool, you need to specify output layers in input_list file. Refer to following reference guide.
https://developer.qualcomm.com/docs/snpe/tools.html#tools_snpe-net-run
Thanks,
Jihoon
yeah the given options proved really helpful at initial stage and you can go for it. Try one o the options for it and it can really prove beneficial for all developers in their Android development.
Yeah, the choices given at the beginning were very helpful and you could make it happen. Try the options for it and all developers in their Android development can really benefit. Visit here for more details.