Model Optimizer Developer Guide

Model Optimizer is a cross-platform command-line tool that facilitates the transition between the training and deployment environment, performs static model analysis, and adjusts deep learning models for optimal execution on end-point target devices.

Model Optimizer process assumes you have a network model trained using a supported deep learning framework. The scheme below illustrates the typical workflow for deploying a trained deep learning model:

Model Optimizer produces an Intermediate Representation (IR) of the network, which can be read, loaded, and inferred with the Inference Engine. The Inference Engine API offers a unified API across a number of supported Intel® platforms. The Intermediate Representation is a pair of files describing the model:

• .xml - Describes the network topology
• .bin - Contains the weights and biases binary data.

TIP: You also can work with the Model Optimizer inside the OpenVINO™ Deep Learning Workbench (DL Workbench). DL Workbench is a platform built upon OpenVINO™ and provides a web-based graphical environment that enables you to optimize, fine-tune, analyze, visualize, and compare performance of deep learning models on various Intel® architecture configurations. In the DL Workbench, you can use most of OpenVINO™ toolkit components.
Proceed to an easy installation from Docker to get started.

What's New in the Model Optimizer in this Release?

• Common changes:
  • Implemented several optimization transformations to replace sub-graphs of operations with HSwish, Mish, Swish and SoftPlus operations.
  • Model Optimizer generates IR keeping shape-calculating sub-graphs by default. Previously, this behavior was triggered if the "--keep_shape_ops" command line parameter was provided. The key is ignored in this release and will be deleted in the next release. To trigger the legacy behavior to generate an IR for a fixed input shape (folding ShapeOf operations and shape-calculating sub-graphs to Constant), use the "--static_shape" command line parameter. Changing model input shape using the Inference Engine API in runtime may fail for such an IR.
  • Fixed Model Optimizer conversion issues resulted in non-reshapeable IR using the Inference Engine reshape API.
  • Enabled transformations to fix non-reshapeable patterns in the original networks:
    • Hardcoded Reshape
      • In Reshape(2D)->MatMul pattern
      • Reshape->Transpose->Reshape when the pattern can be fused to the ShuffleChannels or DepthToSpace operation
    • Hardcoded Interpolate
      • In Interpolate->Concat pattern
    • Added a dedicated requirements file for TensorFlow 2.X as well as the dedicated install prerequisites scripts.
    • Replaced the SparseToDense operation with ScatterNDUpdate-4.
• ONNX*:
  • Enabled an ability to specify the model output tensor name using the "--output" command line parameter.
  • Added support for the following operations:
    • Acosh
    • Asinh
    • Atanh
    • DepthToSpace-11, 13
    • DequantizeLinear-10 (zero_point must be constant)
    • HardSigmoid-1,6
    • QuantizeLinear-10 (zero_point must be constant)
    • ReduceL1-11, 13
    • ReduceL2-11, 13
    • Resize-11, 13 (except mode="nearest" with 5D+ input, mode="tf_crop_and_resize", and attributes exclude_outside and extrapolation_value with non-zero values)
    • ScatterND-11, 13
    • SpaceToDepth-11, 13
• TensorFlow*:
  • Added support for the following operations:
    • Acosh
    • Asinh
    • Atanh
    • CTCLoss
    • EuclideanNorm
    • ExtractImagePatches
    • FloorDiv
• MXNet*:
  • Added support for the following operations:
    • Acosh
    • Asinh
    • Atanh
• Kaldi*:
  • Fixed bug with ParallelComponent support. Now it is fully supported with no restrictions.

NOTE: Intel® System Studio is an all-in-one, cross-platform tool suite, purpose-built to simplify system bring-up and improve system and IoT device application performance on Intel® platforms. If you are using the Intel® Distribution of OpenVINO™ with Intel® System Studio, go to Get Started with Intel® System Studio.

Table of Content

• Preparing and Optimizing your Trained Model with Model Optimizer
  • Configuring Model Optimizer
  • Converting a Model to Intermediate Representation (IR)
    • Converting a Model Using General Conversion Parameters
    • Converting Your Caffe* Model
    • Converting Your TensorFlow* Model
      • Converting BERT from TensorFlow
      • Converting GNMT from TensorFlow
      • Converting YOLO from DarkNet to TensorFlow and then to IR
      • Converting Wide and Deep Models from TensorFlow
      • Converting FaceNet from TensorFlow
      • Converting DeepSpeech from TensorFlow
      • Converting Language Model on One Billion Word Benchmark from TensorFlow
      • Converting Neural Collaborative Filtering Model from TensorFlow*
      • Converting TensorFlow* Object Detection API Models
      • Converting TensorFlow*-Slim Image Classification Model Library Models
      • Converting CRNN Model from TensorFlow*
    • Converting Your MXNet* Model
      • Converting a Style Transfer Model from MXNet
    • Converting Your Kaldi* Model
    • Converting Your ONNX* Model
      • Converting Mask-RCNN ONNX* Model
      • Converting DLRM ONNX* Model
    • Model Optimizations Techniques
    • Cutting parts of the model
    • Sub-graph Replacement in Model Optimizer
      • (Deprecated) Case-Study: Converting SSD models created with the TensorFlow* Object Detection API
      • (Deprecated) Case-Study: Converting Faster R-CNN models created with the TensorFlow* Object Detection API
    • Supported Framework Layers
    • Intermediate Representation and Operation Sets
    • Operations Specification
    • Intermediate Representation suitable for INT8 inference
  • Custom Layers in Model Optimizer
    • Extending Model Optimizer with New Primitives
    • Legacy Mode for Caffe* Custom Layers
  • Model Optimizer Frequently Asked Questions
• Known Issues

Typical Next Step: Preparing and Optimizing your Trained Model with Model Optimizer