[Issue Description ]:
I'm trying to convert faster_rcnn model to dlc format. This model is from official Tensorflow model zoo. Download link: http://download.tensorflow.org/models/object_detection/faster_rcnn_incep... . And I use snpe-tensorflow-to-dlc tool to convert pb file to dlc file. But failed.
[Failure Rate in % ]:100%
[Reproduce Step ]:
system information:
OS Ubuntu 16.04.6 LTS
Python 2.7
SNPE snpe-1.27.1
Tensorflow(CPU) 1.12.0
1. setup SNPE-1.27.1 follow the official guide.(https://developer.qualcomm.com/docs/snpe/setup.html)
2. setup tensorflow1.12.0 (https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-1.12.0-cp...)
3. download the pb files (http://download.tensorflow.org/models/object_detection/faster_rcnn_incep...)
4. use the following command to convert:
~/work/proj/snpe/snpe-sdk/bin/x86_64-linux-clang/snpe-tensorflow-to-dlc --graph frozen_inference_graph.pb --input_dim image_tensor 1,480,853,3 --out_node detection_boxes --out_node detection_scores --out_node detection_classes --out_node num_detections --dlc test2.dl
------- error message ---
2019-09-11 16:22:29.521309: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 AVX512F FMA
2019-09-11 16:22:38.761646: F tensorflow/compiler/jit/deadness_analysis.cc:639] Check failed: it != predicate_map_.end() _SINK
Aborted (core dumped)
[Initial Analysis ]:
It seems that the tensorflow version1.12.0 dont compatible with this pb file. The official guide suggest "model training environment" and "converting environment" should be same, it means I must use the same tensorflow version and python version which i used in my training process. But this page (https://github.com/tensorflow/models/blob/master/research/object_detecti...) says: "... Our frozen inference graphs are generated using the v1.12.0 release version of Tensorflow and we do not guarantee that these will work with other versions ...") . That' OK. We use the same tensorflow version! So I'm puzzled what's going wrong...
Hi, yuanhuayong,
Did you solve this issue? I met same bugs too when trying to convert official ssd_mobilenet_v1 to dlc. Thanks.