Intel® Distribution of OpenVINO™ Toolkit
Community assistance about the Intel® Distribution of OpenVINO™ toolkit, OpenCV, and all aspects of computer vision-related on Intel® platforms.
6300 Discussions

LSTM Support in Openvino | Tensorflow and others.

pankajrawat
Novice
1,679 Views

Does Openvino is fully compatible with RNN layers like LSTM, GRU etc, as openvino is explicitly targeting Computer vision Application.

 

As per official documentation of openvino supported layers.

https://docs.openvinotoolkit.org/latest/openvino_docs_MO_DG_prepare_model_Supported_Frameworks_Layers.html

Tensorflow - We dont find any LSTM, RNN layers mentioned - Below is one post which i raised and currently waiting for inputs

https://community.intel.com/t5/Intel-Distribution-of-OpenVINO/Keras-Tensorflow-backend-Lstm-Runtime-Error/m-p/1202578/highlight/false#M20382

Torch / Onnx- We do find GRU and LSTM mentioned in Onnx Supported layers - Below is one post which i raised and currently waiting for inputs

https://community.intel.com/t5/Intel-Distribution-of-OpenVINO/OpenVino-Pytorch-Error-LSTM/m-p/1201545

Can you please provide working examples and also comment on above posts if possible.

Labels (2)
0 Kudos
1 Solution
Munesh_Intel
Moderator
1,665 Views

Hi Pankaj,

OpenVINO supports LSTM, GRU and RNN.

With regards to Model Optimizer, for ONNX framework, GRU and RNN are fully supported, whilst LSTM is supported with some limitations (Peepholes are not supported).

https://docs.openvinotoolkit.org/latest/openvino_docs_MO_DG_prepare_model_Supported_Frameworks_Layers.html#onnx_supported_operators


And with regards to Inference Engine, more information is available at the following pages:

  • The Inference Engine header file, containing information about layers’ structure:

https://docs.openvinotoolkit.org/2020.4/ie__layers_8h_source.html


  • Variable setting in Inference Engine header file:

https://docs.openvinotoolkit.org/2020.4/ie__layers_8h.html#a4253a83f8170c0b540688fa5081d7947


  • The related Operations Specifications are given below:
    • LSTMCell

https://docs.openvinotoolkit.org/2020.4/openvino_docs_ops_sequence_LSTMCell_1.html


  • LSTMSequence

https://docs.openvinotoolkit.org/2020.4/openvino_docs_ops_sequence_LSTMSequence_1.html


  • GRUCell

https://docs.openvinotoolkit.org/2020.4/openvino_docs_ops_sequence_GRUCell_3.html


  • RNNCell

https://docs.openvinotoolkit.org/2020.4/openvino_docs_ops_sequence_RNNCell_3.html



Regards,

Munesh



View solution in original post

0 Kudos
2 Replies
Munesh_Intel
Moderator
1,666 Views

Hi Pankaj,

OpenVINO supports LSTM, GRU and RNN.

With regards to Model Optimizer, for ONNX framework, GRU and RNN are fully supported, whilst LSTM is supported with some limitations (Peepholes are not supported).

https://docs.openvinotoolkit.org/latest/openvino_docs_MO_DG_prepare_model_Supported_Frameworks_Layers.html#onnx_supported_operators


And with regards to Inference Engine, more information is available at the following pages:

  • The Inference Engine header file, containing information about layers’ structure:

https://docs.openvinotoolkit.org/2020.4/ie__layers_8h_source.html


  • Variable setting in Inference Engine header file:

https://docs.openvinotoolkit.org/2020.4/ie__layers_8h.html#a4253a83f8170c0b540688fa5081d7947


  • The related Operations Specifications are given below:
    • LSTMCell

https://docs.openvinotoolkit.org/2020.4/openvino_docs_ops_sequence_LSTMCell_1.html


  • LSTMSequence

https://docs.openvinotoolkit.org/2020.4/openvino_docs_ops_sequence_LSTMSequence_1.html


  • GRUCell

https://docs.openvinotoolkit.org/2020.4/openvino_docs_ops_sequence_GRUCell_3.html


  • RNNCell

https://docs.openvinotoolkit.org/2020.4/openvino_docs_ops_sequence_RNNCell_3.html



Regards,

Munesh



0 Kudos
Munesh_Intel
Moderator
1,614 Views

Hi Pankaj,


This thread will no longer be monitored since we have provided a solution. If you need any additional information from Intel, please submit a new question.


Regards,

Munesh


0 Kudos
Reply