IntelOpenVINO配置和使⽤
背景
最近的项⽬中，有客户提出在既有的CPU服务器上运⾏CNN的⽅案，即不添加nivida显卡已降低成本。因此调研Intel的OpenVINO⽅案。
OpenVINO是Intel提供的给予卷积神经⽹络的计算机视觉开发包。⽬的在能够快速的在Intel的硬件⽅案上部署和开发计算机视觉⼯程和⽅案。OpenVINO⽀持多种Intel硬件⽅案，包括CPU、集成显卡、Intel Movidius算⼒棒以及FPGA等。
官⽅给出的硬件和系统需求为：
Processors
6th-8th Generation Intel® Core™
Intel® Xeon® v5 family
Intel® Xeon® v6 family
Intel® Pentium® processor N4200/5, N3350/5, N3450/5 with Intel® HD Graphics
Intel Movidius NCS
Operating Systems
Ubuntu* 16.04 long-term support (LTS), 64-bit
CentOS* 7.4 or higher, 64-bit
Yocto Project* Poky Jethro* v2.0.3, 64-bit (for target only)
以上的需求来说⽐较苛刻，可以看出加速模块是在Iris的显卡中，特别是在服务器端E系列的CPU中，基本来说只有E3-v5带有集成显卡的才能⽀持。本⼈的电脑是第四代i7，然⽽如果是采⽤存CPU运⾏的⽅式，⼤可以不必考虑显卡的要求。⽼⼀些的CPU测试也是可以⽀持，但是性能肯定会差。
部署
测试环境 Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz 操作系统 Ubuntu 18.04
然⽽对18.04⽀持并不好，在安装过程中有较多的依赖包在18.04的source中已经被remove，导致依赖环
境检测失败。于是决定直接采⽤docker的⽅式配置16.04LTS的系统。在查docker images的时候意外发现，有⼈已经将配置好Intel OpenVINO的环境上传，欣然接受！
lyh@lyh-All-Series:~$ sudo docker search openvino
[sudo] password for lyh:
NAME                              DESCRIPTION                                    STARS              OFFICIAL            AUTOMATED
cortexica/openvino                Ubuntu 16.04 image with Intel OpenVINO depen (1)
下载image，⼤约需要2.2GB的硬盘空间。运⾏容器：
sudo docker run -it cortexica/openvino /bin/bash
OpenVINO的SDK路径位于：
/opt/intel/
验证环境
运⾏官⽅的Demo程序
root@0bf034642447:/opt/intel/computer_vision_sdk/deployment_tools/demo# ./demo_squeezenet_download_convert_run.sh
脚本会⾃动下载和安装依赖包，并且下载测试Caffe模型将其转成OpenVINO的xml/bin⽂件，测试图⽚是⼀张轿车照⽚，运⾏分类结果如下：
[ INFO ] InferenceEngine:
API version ............ 1.2
Build (13911)
[ INFO ] Parsing input parameters
[ INFO ] Loading plugin
API version ............ 1.2
Build .................. lnx_20180510
Description ....... MKLDNNPlugin
[ INFO ] Loading network files:
/root/openvino_models/ir/squeezenet1.1/l
/root/openvino_models/ir/squeezenet1.1/squeezenet1.1.bin
[ INFO ] Preparing input blobs
[ WARNING ] Image is resized from (787, 259) to (227, 227)
[ INFO ] Batch size is 1
[ INFO ] Preparing output blobs
[ INFO ] Loading model to the plugin
[ INFO ] Starting inference (1 iterations)
[ INFO ] Processing output blobs
Top 10 results:
Image /opt/intel/computer_vision_sdk/deployment_tools/demo/../demo/car.png
817 0.8363345 label sports car, sport car
511 0.0946488 label convertible
479 0.0419131 label car wheel
751 0.0091071 label racer, race car, racing car
436 0.0068161 label beach wagon, station wagon, wagon, estate car, beach waggon, station waggon, waggon
656 0.0037564 label minivan
586 0.0025741 label half track
717 0.0016069 label pickup, pickup truck
864 0.0012027 label tow truck, tow car, wrecker
581 0.0005882 label grille, radiator grille
total inference time: 3.4724500
Average running time of one iteration: 3.4724500 ms
Throughput: 287.9811121 FPS
[ INFO ] Execution successful
Demo程序将图⽚转到227x227⼤⼩，并且进⾏分类。
Tensorflow模型使⽤
实际的项⽬过程中我们的模型是采⽤tensorflow训练的，因此需要将其转换为OpenVINO的xml/bin模式
配置Tensorflow转换环境：
root@0bf034642447:/opt/intel/computer_vision_sdk/deployment_tools/model_optimizer/install_prerequisites# sudo ./install_prerequisites_tf.sh
转换模型：
resized./mo_tf.py --input_meta_graph /home/Project/Iv_v_149/pb_model/a  --output_dir model/
运⾏：
./classification_sample  -d CPU  -i car.png -m /opt/intel/computer_vision_sdk/deployment_tools/model_optimizer/model/l
结果：
[ INFO ] InferenceEngine:
API version ............ 1.2
Build (13911)
[ INFO ] Parsing input parameters
[ INFO ] Loading plugin
API version ............ 1.2
Build .................. lnx_20180510
Description ....... MKLDNNPlugin
[ INFO ] Loading network files:
/opt/intel/computer_vision_sdk/deployment_tools/model_optimizer/model/l
/opt/intel/computer_vision_sdk/deployment_tools/model_optimizer/model/model.ckpt-3800000.bin
[ INFO ] Preparing input blobs
[ WARNING ] Image is resized from (787, 259) to (1920, 1080)
[ INFO ] Input size C: 1 W: 1920 H: 1080
[ INFO ] Batch size is 1
[ INFO ] Preparing output blobs
[ INFO ] Output size: 4
[ INFO ] Output W: 1920 H: 1080
[ INFO ] Loading model to the plugin
[ INFO ] Starting inference (1 iterations)
[ INFO ] Infer total: 482.018
[ INFO ] Processing output blobs
total inference time: 482.0178449
Average running time of one iteration: 482.0178449 ms
Throughput: 2.0746120 FPS
实验来看纯CPU运⾏做图像增强速度还是偏慢，相同情况下采⽤GTX1070可以达到20fps以上。后续需要环境实现OpenVINO在Iris 显卡上运⾏实验。

版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。