https://github.com/pinto0309/20220228_intel_deeplearning_day_hitnet_demo

Special Presentation Demo at Intel IoT Planet 2021 DeepLearning Day / インテル IoT プラネット 2021 DeepLearning Dayの特別講演の発表資料 https://www.intel.co.jp/content/www/jp/ja/now/iot-planet/deep-learning-day.html
https://github.com/pinto0309/20220228_intel_deeplearning_day_hitnet_demo

cuda docker intel onnx openvino

Last synced: 5 months ago
JSON representation

Host: GitHub
URL: https://github.com/pinto0309/20220228_intel_deeplearning_day_hitnet_demo
Owner: PINTO0309
License: apache-2.0
Created: 2022-02-11T01:02:10.000Z (over 3 years ago)
Default Branch: main
Last Pushed: 2022-03-06T06:40:51.000Z (over 3 years ago)
Last Synced: 2025-05-05T22:53:23.452Z (5 months ago)
Topics: cuda, docker, intel, onnx, openvino
Language: Python
Homepage:
Size: 3.43 MB
Stars: 20
Watchers: 2
Forks: 0
Open Issues: 0
Metadata Files:
- Readme: README.md
- License: LICENSE

Awesome Lists containing this project

README

# 20220228_intel_deeplearning_day_hitnet_demo
https://www.intel.co.jp/content/www/jp/ja/now/iot-planet/deep-learning-day.html
## 1. Overview / 概要
This is a demonstration of the steps to convert and infer HITNet, a stereo depth estimation model, using a custom build of OpenVINO.
OpenVINOをカスタムビルドしてステレオ深度推定モデルのHITNetを変換し、推論するまでの手順のデモです。
![132152654-fd689269-537f-4ab1-87fc-b08169311cc7](https://user-images.githubusercontent.com/33194443/153523403-afd059d7-6c5b-496e-a418-4b432e2e0f58.gif)

## 2. Environment / 環境
- Ubuntu 20.04 x86_64
- Docker 20.10.12, build e91ed57
- OpenVINO commit hash: e89db1c6de8eb551949330114d476a2a4be499ed
- ONNX
## 3. Overall flow / 全体の流れ
In order to optimize the process as much as possible, the following processing flow is adopted.
TensorFlow **`pb`** -> TensorFlow **`saved_model`** -> TensorFlow Lite **`tflite`** -> ONNX **`onnx`** -> OpenVINO IR **`xml/bin`**

- [4-1. Procurement of original model .pb / .pb オリジナルモデル.pbの調達](#4-1-procurement-of-original-model--pb-オリジナルモデルpbの調達)
- [4-2. Convert .pb to saved_model / .pbをsaved_modelに変換](#4-2-convert-pb-to-saved_model--pbをsaved_modelに変換)
- [4-3. Convert saved_model to ONNX / saved_modelをONNXに変換](#4-3-convert-saved_model-to-onnx--saved_modelをonnxに変換)
- [4-4. Building OpenVINO / OpenVINOのビルド](#4-4-building-openvino--openvinoのビルド)
- [4-5. Convert ONNX to OpenVINO IR / ONNXをOpenVINO IRへ変換](#4-5-convert-onnx-to-openvino-ir--onnxをopenvino-irへ変換)
- [4-6. Download the Dataset / Datasetのダウンロード](#4-6-download-the-dataset--datasetのダウンロード)
- [4-7. HITNet's ONNX demo / HITNetのONNXデモ](#4-7-hitnets-onnx-demo--hitnetのonnxデモ)
- [4-8. HITNet's OpenVINO demo / HITNetのOpenVINOデモ](#4-8-hitnets-openvino-demo--hitnetのopenvinoデモ)
- [4-9. HITNet's TensorRT demo / HITNetのTensorRTデモ](#4-9-hitnets-tensorrt-demo--hitnetのtensorrtデモ)

## 4. Procedure / 手順
### 4-1. Procurement of original model .pb / オリジナルモデル.pbの調達
Download the official HITNet model published by Google Research [here](https://github.com/google-research/google-research/tree/master/hitnet). The file to be downloaded is a Protocol Buffers format file.
[こちら](https://github.com/google-research/google-research/tree/master/hitnet) のGoogle Researchが公開しているHITNet公式モデルをダウンロードします。ダウンロードするファイルはProtocol Buffers形式のファイルです。
```bash
$ git clone https://github.com/PINTO0309/20210228_intel_deeplearning_day_hitnet_demo
$ cd 20210228_intel_deeplearning_day_hitnet_demo

# [1, ?, ?, 2], Grayscale image x2
$ wget https://storage.googleapis.com/tensorflow-graphics/models/hitnet/default_models/eth3d.pb
or
# [1, ?, ?, 6], RGB image x2
$ wget https://storage.googleapis.com/tensorflow-graphics/models/hitnet/default_models/flyingthings_finalpass_xl.pb
or
# [1, ?, ?, 6], RGB image x2
$ wget https://storage.googleapis.com/tensorflow-graphics/models/hitnet/default_models/middlebury_d400.pb
```
Use [Netron](https://netron.app/) to check the structure of the model. In the case of eth3d, two grayscale images of one channel are used as input. The name of the input is **`input`**.
モデルの構造を確認するには、[Netron](https://netron.app/) を使用します。eth3dの場合、1チャンネルのグレースケール画像2枚を入力として使用します。入力の名前は **`input`** です。
![image](https://user-images.githubusercontent.com/33194443/153540670-354a575c-2c0a-4f1f-b350-767bfb2b1e5d.png)
The name of the output is **`reference_output_disparity`**.
出力の名前は **`reference_output_disparity`** です。
![image](https://user-images.githubusercontent.com/33194443/153558437-da09fa51-aa84-4ecc-b42c-4197b8c06281.png)

For non-eth3d, the input is two 3-channel RGB images.
eth3d以外のモデルの場合、入力は3チャンネルのRGB画像2枚です。
![image](https://user-images.githubusercontent.com/33194443/153541985-7e3e580d-b659-4532-b0e3-28bc2fea0957.png)
[↥ Back to top](#3-overall-flow--全体の流れ)
### 4-2. Convert .pb to saved_model / .pbをsaved_modelに変換
Start a Docker container with all the latest versions of the various major frameworks such as OpenVINO, TensorFlow, PyTorch, ONNX, etc. Note that the Docker Image is quite large, 26GB, since all the huge frameworks such as CUDA and TensorRT are also installed. Also, in order to launch the demo with GUI from within the Docker container, many launch options are specified, such as **`xhost`**, **`--gpus`**, **`-v`**, **`-e`**, **`--net`**, **`--privileged`**, etc., but they do not need to be specified if you do not want to use the GUI. If you want to know what kind of framework is implemented in a Docker container, please click [here](https://github.com/PINTO0309/openvino2tensorflow#1-environment).
OpenVINOやTensorFlowやPyTorchやONNXなどの各種主要フレームワークの最新バージョンが全て導入されたDockerコンテナを起動します。CUDAやTensorRTなどの巨大なフレームワークも全てインストールされているため、Docker Imageは26GBとかなり大きいことに注意してください。また、Dockerコンテナの中からGUIを使用したデモを起動するため、**`xhost`**, **`--gpus`**, **`-v`**, **`-e`**, **`--net`**, **`--privileged`** などの多くの起動オプションを指定していますが、GUIを使用しない場合は指定不要です。どのようなフレームワークが導入されたDockerコンテナかを知りたい場合は [こちら](https://github.com/PINTO0309/openvino2tensorflow#1-environment) をご覧ください。
```bash
$ xhost +local: && \
docker run --gpus all -it --rm \
-v `pwd`:/home/user/workdir \
-v /tmp/.X11-unix/:/tmp/.X11-unix:rw \
--net=host \
-e XDG_RUNTIME_DIR=$XDG_RUNTIME_DIR \
-e DISPLAY=$DISPLAY \
--privileged \
ghcr.io/pinto0309/openvino2tensorflow:latest
```
```bash
$ MODEL=eth3d
or
$ MODEL=flyingthings_finalpass_xl
or
$ MODEL=middlebury_d400

$ pb_to_saved_model \
--pb_file_path ${MODEL}.pb \
--inputs input:0 \
--outputs reference_output_disparity:0 \
--model_output_path ${MODEL}/saved_model
```
A sample without GUI is shown below.
GUIを使用しない場合のサンプルは下記のとおりです。
```bash
$ docker run -it --rm \
-v `pwd`:/home/user/workdir \
ghcr.io/pinto0309/openvino2tensorflow:latest
```
```bash
$ MODEL=eth3d
or
$ MODEL=flyingthings_finalpass_xl
or
$ MODEL=middlebury_d400

$ pb_to_saved_model \
--pb_file_path ${MODEL}.pb \
--inputs input:0 \
--outputs reference_output_disparity:0 \
--model_output_path ${MODEL}/saved_model
```
Let's check the shape of the generated **`saved_model`**, using the standard TensorFlow tool **`saved_model_cli`**.Of the input NHWC shape **`batch,height,width,channel`**, the height and width are undefined **`-1`**.
生成された **`saved_model`** の形状を確認してみます。TensorFlowの標準ツール **`saved_model_cli`** を使用します。入力のNHWC形状 **`バッチサイズ,高さ,幅,チャンネル`** のうち、高さと幅が未定義の **`-1`** となっています。
```bash
$ saved_model_cli show --dir middlebury_d400/saved_model/ --all

MetaGraphDef with tag-set: 'serve' contains the following SignatureDefs:

signature_def['serving_default']:
The given SavedModel SignatureDef contains the following input(s):
inputs['input'] tensor_info:
dtype: DT_FLOAT
shape: (-1, -1, -1, 6)
name: input:0
The given SavedModel SignatureDef contains the following output(s):
outputs['reference_output_disparity'] tensor_info:
dtype: DT_FLOAT
shape: (-1, -1, -1, -1)
name: reference_output_disparity:0
Method name is: tensorflow/serving/predict
```
[↥ Back to top](#3-overall-flow--全体の流れ)
### 4-3. Convert saved_model to ONNX / saved_modelをONNXに変換
The tool **`saved_model_to_tflite`** introduced in the Dokcer container is used to generate **`tflite`** from **`saved_model`**. The tool **`tensorflow-onnx`** can be used to generate **`onnx`** from **`saved_model`** immediately, but I will convert it once to **`tflite`** to make it as optimized as possible. The **`--input_shapes`** option can be used to fix undefined input shapes to a specified size.
Dokcerコンテナに導入されている **`saved_model_to_tflite`** というツールを使用して **`saved_model`** から **`tflite`** を生成します。公式の **`tensorflow-onnx`** というツールを使用すると **`saved_model`** から即座に **`onnx`** を生成することが可能ですが、なるべく最適化を行うためにあえて一度 **`tflite`** へ変換します。**`--input_shapes`** オプションを使用することで未定義の入力形状を指定のサイズへ固定することができます。
```bash
$ H=480
$ W=640
$ saved_model_to_tflite \
--saved_model_dir_path ${MODEL}/saved_model \
--input_shapes [1,${H},${W},6] \
--model_output_dir_path ${MODEL}/saved_model_${H}x${W} \
--output_no_quant_float32_tflite
```
Check the input and output structure of the generated TFLite. At this point, TensorFlowLite's optimizer has already removed a large number of unnecessary operations or merged multiple operations into a clean and simple structure.
生成されたTFLiteの入力と出力の構造を確認します。この時点ですでにTensorFlowLiteのオプティマイザによって不要なオペレーションが大量に削除されたり、あるいは複数のオペレーションが融合して綺麗でシンプルな構造に変換されています。
![image](https://user-images.githubusercontent.com/33194443/153568491-4165d552-fa2e-4fa8-ae3e-aa288b7998cc.png)
Next, convert **`tflite`** to **`onnx`**. I will use **`tensorflow-onnx`** here. **`--inputs-as-nchw input`** is an option to convert the shape of the input from **`NHWC`** to **`NCHW`**. Note that the onnx opset to be generated must be **`12`**.
次に、**`tflite`** を **`onnx`** へ変換します。ここで **`tensorflow-onnx`** を使用します。**`--inputs-as-nchw input`** は入力の形状を **`NHWC`** から **`NCHW`** へ変換するためのオプションです。なお、生成するonnxのopsetは **`12`** を指定する必要があります。
```bash
$ python -m tf2onnx.convert \
--opset 12 \
--inputs-as-nchw input \
--tflite ${MODEL}/saved_model_${H}x${W}/model_float32.tflite \
--output ${MODEL}/saved_model_${H}x${W}/model_float32.onnx
```
Redundant onnx files are generated with insufficient optimization and undefined input/output information for each operation.
最適化が不十分で、なおかつ各オペレーションの入出力情報が未定義の冗長なonnxファイルが生成されます。
![image](https://user-images.githubusercontent.com/33194443/153574176-32dd914a-47e9-46b6-9d9a-a2ebaa2c52c8.png)
Uses **`onnx-simplifier`** to further optimize onnx files.
**`onnx-simplifier`** を使用してonnxファイルをさらに最適化します。
```bash
$ python -m onnxsim \
${MODEL}/saved_model_${H}x${W}/model_float32.onnx \
${MODEL}/saved_model_${H}x${W}/model_float32.onnx
```
The file size will increase, but the structure of the model will be optimized and inference performance will not be affected.
ファイルサイズが肥大化しますが、モデルの構造は最適化されおり推論パフォーマンスに影響はありません。
![image](https://user-images.githubusercontent.com/33194443/153575177-3b9c5b06-080b-45fc-bfbb-c4814b5ac00d.png)
[↥ Back to top](#3-overall-flow--全体の流れ)
### 4-4. Building OpenVINO / OpenVINOのビルド
Since there are some issues with the current latest version of the OpenVINO model optimizer, we will build OpenVINO itself from the source code of the commits that have already resolved the [issues](github.com/openvinotoolkit/openvino/issues/7379).
OpenVINOモデルオプティマイザの現行最新バージョンには一部問題があるため、問題箇所を解消済みのコミットのソースコードからOpenVINOそのものをビルドします。Intelのエンジニアとやりとりして解消いただいた問題点の内容が気になる方は [こちら](https://github.com/openvinotoolkit/openvino/issues/7379) をご覧ください。
```bash
$ git clone https://github.com/openvinotoolkit/openvino \
&& cd openvino \
&& git checkout e89db1c6de8eb551949330114d476a2a4be499ed \
&& git submodule update --init --recursive \
&& pip install pip --upgrade \
&& pip install Cython numpy setuptools wheel pafy youtube-dl \
&& chmod +x scripts/submodule_update_with_gitee.sh \
&& ./scripts/submodule_update_with_gitee.sh \
&& chmod +x install_build_dependencies.sh \
&& ./install_build_dependencies.sh \
&& mkdir build \
&& cd build \
&& cmake \
-DCMAKE_BUILD_TYPE=Release \
-DENABLE_PYTHON=ON \
-DPYTHON_EXECUTABLE=`which python3` \
-DPYTHON_LIBRARY=/usr/lib/x86_64-linux-gnu/libpython3.8.so \
-DPYTHON_INCLUDE_DIR=/usr/include/python3.8 \
-DENABLE_CLDNN=ON \
-DENABLE_WHEEL=ON .. \
&& make -j$(nproc)
```
Build finished.
ビルド終了。
![image](https://user-images.githubusercontent.com/33194443/153585828-6f5575da-c0b4-42f7-a3a9-57f3cf8840e5.png)
Check the generated Wheel files; two Wheel files have been generated.
生成されたWheelファイルを確認します。Wheelファイルは２個生成されています。
```bash
$ ls -l wheels/*
-rw-r--r-- 1 user user 30777895 Feb 11 11:17 wheels/openvino-2022.1.0-000-cp38-cp38-manylinux_2_31_x86_64.whl
-rw-r--r-- 1 user user 6419721 Feb 11 11:06 wheels/openvino_dev-2022.1.0-000-py3-none-any.whl
```
Overwrite the OpenVINO installation.
OpenVINOを上書きインストールします。
```bash
$ sudo ${INTEL_OPENVINO_DIR}/openvino_toolkit_uninstaller/uninstall.sh --silent \
&& sudo pip install wheels/* && cd ../.. && rm -rf openvino
```
[↥ Back to top](#3-overall-flow--全体の流れ)
### 4-5. Convert ONNX to OpenVINO IR / ONNXをOpenVINO IRへ変換
Convert ONNX files to OpenVINO IR.
ONNXファイルをOpenVINO IRへ変換します。
```bash
$ sudo python /usr/local/lib/python3.8/dist-packages/openvino/tools/mo/mo.py \
--input_model ${MODEL}/saved_model_${H}x${W}/model_float32.onnx \
--data_type FP32 \
--output_dir ${MODEL}/saved_model_${H}x${W}/openvino/FP32 \
--model_name ${MODEL}_${H}x${W} \
&& sudo chown -R user ${MODEL}
```
```console
/usr/local/lib/python3.8/dist-packages/pkg_resources/__init__.py:122: PkgResourcesDeprecationWarning: 0.1.9-nvc is an invalid version and will not be supported in a future release
warnings.warn(
/usr/local/lib/python3.8/dist-packages/pkg_resources/__init__.py:122: PkgResourcesDeprecationWarning: 0.1.9-nvc is an invalid version and will not be supported in a future release
warnings.warn(
Model Optimizer arguments:
Common parameters:
- Path to the Input Model: /home/user/workdir/middlebury_d400/saved_model_480x640/model_float32.onnx
- Path for generated IR: /home/user/workdir/middlebury_d400/saved_model_480x640/openvino/FP32
- IR output name: middlebury_d400_480x640
- Log level: ERROR
- Batch: Not specified, inherited from the model
- Input layers: Not specified, inherited from the model
- Output layers: Not specified, inherited from the model
- Input shapes: Not specified, inherited from the model
- Source layout: Not specified
- Target layout: Not specified
- Layout: Not specified
- Mean values: Not specified
- Scale values: Not specified
- Scale factor: Not specified
- Precision of IR: FP32
- Enable fusing: True
- Enable grouped convolutions fusing: True
- Move mean values to preprocess section: None
- Reverse input channels: False
- Use legacy API for model processing: False
- Use the transformations config file: None
ONNX specific parameters:
/usr/local/lib/python3.8/dist-packages/pkg_resources/__init__.py:122: PkgResourcesDeprecationWarning: 0.1.9-nvc is an invalid version and will not be supported in a future release
warnings.warn(
- OpenVINO runtime found in: /usr/local/lib/python3.8/dist-packages/openvino
OpenVINO runtime version: 2022.1.custom_HEAD_e89db1c6de8eb551949330114d476a2a4be499ed
Model Optimizer version: custom_HEAD_e89db1c6de8eb551949330114d476a2a4be499ed
[ WARNING ] Model Optimizer and OpenVINO runtime versions do no match.
[ WARNING ] Consider building the OpenVINO Python API from sources or reinstall OpenVINO (TM) toolkit using "pip install openvino" (may be incompatible with the current Model Optimizer version)
[ WARNING ]
Detected not satisfied dependencies:
numpy: installed: 1.22.2, required: < 1.20
fastjsonschema: not installed, required: ~= 2.15.1

Please install required versions of components or use install_prerequisites script
/usr/local/lib/python3.8/dist-packages/openvino/tools/mo/install_prerequisites/install_prerequisites_onnx.sh
Note that install_prerequisites scripts may install additional components.
/usr/local/lib/python3.8/dist-packages/pkg_resources/__init__.py:122: PkgResourcesDeprecationWarning: 0.1.9-nvc is an invalid version and will not be supported in a future release
warnings.warn(
[ SUCCESS ] Generated IR version 11 model.
[ SUCCESS ] XML file: /home/user/workdir/middlebury_d400/saved_model_480x640/openvino/FP32/middlebury_d400_480x640.xml
[ SUCCESS ] BIN file: /home/user/workdir/middlebury_d400/saved_model_480x640/openvino/FP32/middlebury_d400_480x640.bin
[ SUCCESS ] Total execution time: 50.17 seconds.
[ SUCCESS ] Memory consumed: 410 MB.
```
![image](https://user-images.githubusercontent.com/33194443/153590231-fe9f9ddd-6e1e-4189-8db4-40a8f5fa950f.png)
[↥ Back to top](#3-overall-flow--全体の流れ)
### 4-6. Download the Dataset / Datasetのダウンロード
Download a stereo driving dataset for testing. It is hard to see, but it is a dataset of pairs of images taken from each of the two left and right cameras.
テスト用のステレオドライビングデータセットをダウンロードします。見た目では分かりにくいですが、２個の左右のカメラからそれぞれ撮影した画像のペアのデータセットです。
|Left|Right|
|:--:|:--:|
|![2018-07-11-14-48-52_2018-07-11-14-50-08-769_L](https://user-images.githubusercontent.com/33194443/153627778-0e433bab-7509-4b50-a77e-18596416ae02.jpg)|![2018-07-11-14-48-52_2018-07-11-14-50-08-769_R](https://user-images.githubusercontent.com/33194443/153627817-58af4544-47f9-4dca-8a32-21826491c92e.jpg)|
```bash
$ mkdir -p "DrivingStereo images/left" \
&& mkdir -p "DrivingStereo images/right" \
&& mkdir -p "DrivingStereo images/depth" \
&& wget https://github.com/PINTO0309/20210228_intel_deeplearning_day_hitnet_demo/releases/download/v1.0/2018-07-11-14-48-52_left.zip \
&& unzip -d "DrivingStereo images/left" -q 2018-07-11-14-48-52_left.zip \
&& rm 2018-07-11-14-48-52_left.zip \
&& wget https://github.com/PINTO0309/20210228_intel_deeplearning_day_hitnet_demo/releases/download/v1.0/2018-07-11-14-48-52_right.zip \
&& unzip -d "DrivingStereo images/right" -q 2018-07-11-14-48-52_right.zip \
&& rm 2018-07-11-14-48-52_right.zip \
&& wget https://github.com/PINTO0309/20210228_intel_deeplearning_day_hitnet_demo/releases/download/v1.0/2018-07-11-14-48-52_depth.zip \
&& unzip -d "DrivingStereo images/depth" -q 2018-07-11-14-48-52_depth.zip \
&& rm 2018-07-11-14-48-52_depth.zip \
&& wget https://github.com/PINTO0309/20210228_intel_deeplearning_day_hitnet_demo/releases/download/v1.0/stereo_movie.mp4
```
### 4-7. HITNet's ONNX demo / HITNetのONNXデモ
#### 4-7-1. ONNX+CUDA
I'll borrow ibaiGorordo's ONNX demo to run it. Adjust the program slightly so that ONNX's CUDA provider is enabled.
ibaiGorordoさんのONNXデモをお借りして実行してみます。ONNXのCUDAプロバイダが有効になるように、プログラムを少しだけ調整します。
```bash
$ rm -rf ONNX-HITNET-Stereo-Depth-estimation \
&& git clone https://github.com/ibaiGorordo/ONNX-HITNET-Stereo-Depth-estimation.git \
&& cd ONNX-HITNET-Stereo-Depth-estimation \
&& git checkout 20471bfe2a23c34681141a9c0401eeff45680330 \
&& cd .. \
&& sed -i 's/models\///g' ONNX-HITNET-Stereo-Depth-estimation/drivingStereoTest.py \
&& sed -i 's/cv2.WINDOW_NORMAL/cv2.WINDOW_AUTOSIZE/g' ONNX-HITNET-Stereo-Depth-estimation/drivingStereoTest.py \
&& sed -i 's/max_distance = 30/max_distance = 80/g' ONNX-HITNET-Stereo-Depth-estimation/drivingStereoTest.py \
&& sed -i 's/np.hstack((left_img,color_real_depth, color_depth))/np.hstack((left_img, color_depth))/g' ONNX-HITNET-Stereo-Depth-estimation/drivingStereoTest.py \
&& sed -i '31i \\t\tsession_option = onnxruntime.SessionOptions()' ONNX-HITNET-Stereo-Depth-estimation/hitnet/hitnet.py \
&& sed -i '32i \\t\tmodel_file_name = model_path.split(".")[0]' ONNX-HITNET-Stereo-Depth-estimation/hitnet/hitnet.py \
&& sed -i '33i \\t\tsession_option.optimized_model_filepath = f"{model_file_name}_cudaopt.onnx"' ONNX-HITNET-Stereo-Depth-estimation/hitnet/hitnet.py \
&& sed -i '34i \\t\tsession_option.graph_optimization_level = onnxruntime.GraphOptimizationLevel.ORT_ENABLE_EXTENDED' ONNX-HITNET-Stereo-Depth-estimation/hitnet/hitnet.py \
&& sed -i "s/onnxruntime.InferenceSession(model_path/onnxruntime.InferenceSession(model_path, session_option, providers=[\'CUDAExecutionProvider\']/g" ONNX-HITNET-Stereo-Depth-estimation/hitnet/hitnet.py

$ python ONNX-HITNET-Stereo-Depth-estimation/drivingStereoTest.py
```
![image](https://user-images.githubusercontent.com/33194443/153699128-b012dab9-5dcb-44d7-9750-56f913982543.png)
#### 4-7-2. ONNX+TensorRT
```bash
$ rm -rf ONNX-HITNET-Stereo-Depth-estimation \
&& git clone https://github.com/ibaiGorordo/ONNX-HITNET-Stereo-Depth-estimation.git \
&& cd ONNX-HITNET-Stereo-Depth-estimation \
&& git checkout 20471bfe2a23c34681141a9c0401eeff45680330 \
&& cd .. \
&& sed -i 's/models\///g' ONNX-HITNET-Stereo-Depth-estimation/drivingStereoTest.py \
&& sed -i 's/cv2.WINDOW_NORMAL/cv2.WINDOW_AUTOSIZE/g' ONNX-HITNET-Stereo-Depth-estimation/drivingStereoTest.py \
&& sed -i 's/max_distance = 30/max_distance = 80/g' ONNX-HITNET-Stereo-Depth-estimation/drivingStereoTest.py \
&& sed -i 's/np.hstack((left_img,color_real_depth, color_depth))/np.hstack((left_img, color_depth))/g' ONNX-HITNET-Stereo-Depth-estimation/drivingStereoTest.py \
&& sed -i "s/onnxruntime.InferenceSession(model_path/onnxruntime.InferenceSession(model_path, providers=[\'TensorrtExecutionProvider', 'CUDAExecutionProvider']/g" ONNX-HITNET-Stereo-Depth-estimation/hitnet/hitnet.py
```
https://user-images.githubusercontent.com/33194443/156153624-4a94754e-bfaf-470f-a830-2c9483efb474.mp4

[↥ Back to top](#3-overall-flow--全体の流れ)
### 4-8. HITNet's OpenVINO demo / HITNetのOpenVINOデモ
Run a test inference program customized for OpenVINO: CPU inference.
OpenVINO用にカスタマイズしたテスト用推論プログラムを実行します。CPU推論です。
```bash
$ python drivingStereoTest_openvino.py
```
![image](https://user-images.githubusercontent.com/33194443/153699261-ef2e1cd2-eae4-4568-825c-faa2d61b125d.png)
[↥ Back to top](#3-overall-flow--全体の流れ)
### 4-9. HITNet's TensorRT demo / HITNetのTensorRTデモ
I will be borrowing iwatake's TensorRT demo to run the test. Follow the tutorial in this repository to set up and run the environment.
iwatakeさんのTensorRTデモをお借りしてテストを実施します。こちらのリポジトリのチュートリアルに従って環境を構築して実行します。
https://github.com/iwatake2222/play_with_tensorrt/tree/master/pj_tensorrt_depth_stereo_hitnet
```bash
$ ./main stereo_movie.mp4
```
![image](https://user-images.githubusercontent.com/33194443/153711552-6bcb17f8-90d5-4526-9382-0684c9d79dcf.png)
[↥ Back to top](#3-overall-flow--全体の流れ)
## 5. Acknowledgements / 謝辞
Thanks!!!
- **Intel Team**:
- [GatherND shape conversion from ONNX is inaccurate #7379](https://github.com/openvinotoolkit/openvino/issues/7379)
- [Const data got different desc and content byte sizes (24 and 96 respectively)" error when converting ConvolutionBackpropData using compile_tools #9517](https://github.com/openvinotoolkit/openvino/issues/9517)
- **[openvinotoolkit](https://github.com/openvinotoolkit)**:
- https://github.com/openvinotoolkit/openvino

LICENSE

```
Apache License
Version 2.0, January 2004
http://www.apache.org/licenses/

TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION

1. Definitions.

"License" shall mean the terms and conditions for use, reproduction,
and distribution as defined by Sections 1 through 9 of this document.

"Licensor" shall mean the copyright owner or entity authorized by
the copyright owner that is granting the License.

"Legal Entity" shall mean the union of the acting entity and all
other entities that control, are controlled by, or are under common
control with that entity. For the purposes of this definition,
"control" means (i) the power, direct or indirect, to cause the
direction or management of such entity, whether by contract or
otherwise, or (ii) ownership of fifty percent (50%) or more of the
outstanding shares, or (iii) beneficial ownership of such entity.

"You" (or "Your") shall mean an individual or Legal Entity
exercising permissions granted by this License.

"Source" form shall mean the preferred form for making modifications,
including but not limited to software source code, documentation
source, and configuration files.

"Object" form shall mean any form resulting from mechanical
transformation or translation of a Source form, including but
not limited to compiled object code, generated documentation,
and conversions to other media types.

"Work" shall mean the work of authorship, whether in Source or
Object form, made available under the License, as indicated by a
copyright notice that is included in or attached to the work
(an example is provided in the Appendix below).

"Derivative Works" shall mean any work, whether in Source or Object
form, that is based on (or derived from) the Work and for which the
editorial revisions, annotations, elaborations, or other modifications
represent, as a whole, an original work of authorship. For the purposes
of this License, Derivative Works shall not include works that remain
separable from, or merely link (or bind by name) to the interfaces of,
the Work and Derivative Works thereof.

"Contribution" shall mean any work of authorship, including
the original version of the Work and any modifications or additions
to that Work or Derivative Works thereof, that is intentionally
submitted to Licensor for inclusion in the Work by the copyright owner
or by an individual or Legal Entity authorized to submit on behalf of
the copyright owner. For the purposes of this definition, "submitted"
means any form of electronic, verbal, or written communication sent
to the Licensor or its representatives, including but not limited to
communication on electronic mailing lists, source code control systems,
and issue tracking systems that are managed by, or on behalf of, the
Licensor for the purpose of discussing and improving the Work, but
excluding communication that is conspicuously marked or otherwise
designated in writing by the copyright owner as "Not a Contribution."

"Contributor" shall mean Licensor and any individual or Legal Entity
on behalf of whom a Contribution has been received by Licensor and
subsequently incorporated within the Work.

2. Grant of Copyright License. Subject to the terms and conditions of
this License, each Contributor hereby grants to You a perpetual,
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
copyright license to reproduce, prepare Derivative Works of,
publicly display, publicly perform, sublicense, and distribute the
Work and such Derivative Works in Source or Object form.

3. Grant of Patent License. Subject to the terms and conditions of
this License, each Contributor hereby grants to You a perpetual,
worldwide, non-exclusive, no-charge, royalty-free, irrevocable
(except as stated in this section) patent license to make, have made,
use, offer to sell, sell, import, and otherwise transfer the Work,
where such license applies only to those patent claims licensable
by such Contributor that are necessarily infringed by their
Contribution(s) alone or by combination of their Contribution(s)
with the Work to which such Contribution(s) was submitted. If You
institute patent litigation against any entity (including a
cross-claim or counterclaim in a lawsuit) alleging that the Work
or a Contribution incorporated within the Work constitutes direct
or contributory patent infringement, then any patent licenses
granted to You under this License for that Work shall terminate
as of the date such litigation is filed.

4. Redistribution. You may reproduce and distribute copies of the
Work or Derivative Works thereof in any medium, with or without
modifications, and in Source or Object form, provided that You
meet the following conditions:

(a) You must give any other recipients of the Work or
Derivative Works a copy of this License; and

(b) You must cause any modified files to carry prominent notices
stating that You changed the files; and

(c) You must retain, in the Source form of any Derivative Works
that You distribute, all copyright, patent, trademark, and
attribution notices from the Source form of the Work,
excluding those notices that do not pertain to any part of
the Derivative Works; and

(d) If the Work includes a "NOTICE" text file as part of its
distribution, then any Derivative Works that You distribute must
include a readable copy of the attribution notices contained
within such NOTICE file, excluding those notices that do not
pertain to any part of the Derivative Works, in at least one
of the following places: within a NOTICE text file distributed
as part of the Derivative Works; within the Source form or
documentation, if provided along with the Derivative Works; or,
within a display generated by the Derivative Works, if and
wherever such third-party notices normally appear. The contents
of the NOTICE file are for informational purposes only and
do not modify the License. You may add Your own attribution
notices within Derivative Works that You distribute, alongside
or as an addendum to the NOTICE text from the Work, provided
that such additional attribution notices cannot be construed
as modifying the License.

You may add Your own copyright statement to Your modifications and
may provide additional or different license terms and conditions
for use, reproduction, or distribution of Your modifications, or
for any such Derivative Works as a whole, provided Your use,
reproduction, and distribution of the Work otherwise complies with
the conditions stated in this License.

5. Submission of Contributions. Unless You explicitly state otherwise,
any Contribution intentionally submitted for inclusion in the Work
by You to the Licensor shall be under the terms and conditions of
this License, without any additional terms or conditions.
Notwithstanding the above, nothing herein shall supersede or modify
the terms of any separate license agreement you may have executed
with Licensor regarding such Contributions.

6. Trademarks. This License does not grant permission to use the trade
names, trademarks, service marks, or product names of the Licensor,
except as required for reasonable and customary use in describing the
origin of the Work and reproducing the content of the NOTICE file.

7. Disclaimer of Warranty. Unless required by applicable law or
agreed to in writing, Licensor provides the Work (and each
Contributor provides its Contributions) on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
implied, including, without limitation, any warranties or conditions
of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
PARTICULAR PURPOSE. You are solely responsible for determining the
appropriateness of using or redistributing the Work and assume any
risks associated with Your exercise of permissions under this License.

8. Limitation of Liability. In no event and under no legal theory,
whether in tort (including negligence), contract, or otherwise,
unless required by applicable law (such as deliberate and grossly
negligent acts) or agreed to in writing, shall any Contributor be
liable to You for damages, including any direct, indirect, special,
incidental, or consequential damages of any character arising as a
result of this License or out of the use or inability to use the
Work (including but not limited to damages for loss of goodwill,
work stoppage, computer failure or malfunction, or any and all
other commercial damages or losses), even if such Contributor
has been advised of the possibility of such damages.

9. Accepting Warranty or Additional Liability. While redistributing
the Work or Derivative Works thereof, You may choose to offer,
and charge a fee for, acceptance of support, warranty, indemnity,
or other liability obligations and/or rights consistent with this
License. However, in accepting such obligations, You may act only
on Your own behalf and on Your sole responsibility, not on behalf
of any other Contributor, and only if You agree to indemnify,
defend, and hold each Contributor harmless for any liability
incurred by, or claims asserted against, such Contributor by reason
of your accepting any such warranty or additional liability.

END OF TERMS AND CONDITIONS

APPENDIX: How to apply the Apache License to your work.

To apply the Apache License to your work, attach the following
boilerplate notice, with the fields enclosed by brackets "[]"
replaced with your own identifying information. (Don't include
the brackets!) The text should be enclosed in the appropriate
comment syntax for the file format. We also recommend that a
file or class name and description of purpose be included on the
same "printed page" as the copyright notice for easier
identification within third-party archives.

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
```

- **[NobuoTsukamoto](https://github.com/NobuoTsukamoto)**:
- https://github.com/NobuoTsukamoto/benchmarks

LICENSE

```
MIT License

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.
```

- **[ibaiGorordo](https://github.com/ibaiGorordo)**:
- https://github.com/ibaiGorordo/ONNX-HITNET-Stereo-Depth-estimation

LICENSE

```
MIT License

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

- **[iwatake2222](https://github.com/iwatake2222)**:
- https://github.com/iwatake2222/play_with_tensorrt

LICENSE

```
Apache License
Version 2.0, January 2004
http://www.apache.org/licenses/

TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION

1. Definitions.

"License" shall mean the terms and conditions for use, reproduction,
and distribution as defined by Sections 1 through 9 of this document.

"Licensor" shall mean the copyright owner or entity authorized by
the copyright owner that is granting the License.

"You" (or "Your") shall mean an individual or Legal Entity
exercising permissions granted by this License.

"Source" form shall mean the preferred form for making modifications,
including but not limited to software source code, documentation
source, and configuration files.

"Contributor" shall mean Licensor and any individual or Legal Entity
on behalf of whom a Contribution has been received by Licensor and
subsequently incorporated within the Work.

(a) You must give any other recipients of the Work or
Derivative Works a copy of this License; and

(b) You must cause any modified files to carry prominent notices
stating that You changed the files; and

END OF TERMS AND CONDITIONS

APPENDIX: How to apply the Apache License to your work.

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

- **A Large-Scale Dataset for Stereo Matching in Autonomous Driving Scenarios**:
- https://drivingstereo-dataset.github.io/

LICENSE

```
MIT License

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

```
@inproceedings{yang2019drivingstereo,
title={DrivingStereo: A Large-Scale Dataset for Stereo Matching in Autonomous Driving Scenarios},
author={Yang, Guorun and Song, Xiao and Huang, Chaoqin and Deng, Zhidong and Shi, Jianping and Zhou, Bolei},
booktitle={IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
year={2019}
}
```

ecosyste.ms

Data

Tools

Indexes

Applications

Experiments

Awesome

https://github.com/pinto0309/20220228_intel_deeplearning_day_hitnet_demo

Awesome Lists containing this project

README