在京东预定了好几天,可惜到了几周,总是加班太忙,今天总算尝了鲜。
win10_x64,已安装VS2015 + Python3.6(Anaconda3) + CMake3.12.3
还可以安装OpenCV4.0.0,openvino里面的demo都用到了opencv。但这些demo都是用cmake编译的,cmake加载的opencv路径,我始终没有配置成功。所以,最好是使用openvino安装opencv环境,来进行demo编译。
下载w_openvino_toolkit_p_2018.4.420.exe,安装以下选项:
[img]http://dl2.iteye.com/upload/attachment/0131/2199/a9e2e82f-f4c9-32ae-8f13-0327636fbd91.jpg" alt="[/img]
算法是可以CPU跑的,咱要测试VPU,最好选上OpenCV。
安装完毕之后。我还以为有IDE工具呢,想多了。接着要做几件事:
1,启动CMD,执行:
D:\Intel\computer_vision_sdk\deployment_tools\model_optimizer\install_prerequisites\install_prerequisites.bat
如果明确只使用一种架构,比如caffe,那么可以只执行:install_prerequisites_caffe.bat
2,添加环境变量:
D:\Intel\computer_vision_sdk_2018.4.420\inference_engine\bin\intel64\Release
否则运行demo时,会出现找不到inference_engine.dll的错误。
3,用管理员身份启动CMD,执行:
D:\Intel\computer_vision_sdk_2018.4.420\deployment_tools\demo\demo_squeezenet_download_convert_run -d MYRIAD
注意这里,一定要加“-d MYRIAD”,是指运行在什么设备上,否则之后编译出来的模型是FP32的,而VPU只支持FP16。
上面的批处理跑到最后,会执行失败,那么就去
D:\Intel\computer_vision_sdk_2018.4.420\deployment_tools\inference_engine\samples\intel64\Release
找到编译好的可执行文件:classification_sample.exe
执行指令:
classification_sample.exe -i "D:\Intel\computer_vision_sdk_2018.4.420\deployment_tools\demo\car.png" -m "C:\Users\ADong\Documents\Intel\OpenVINO\models\ir\squeezenet1.1.xml" -d MYRIAD
算力还是相当牛掰。
用VS2015打开:
D:\Intel\computer_vision_sdk_2018.4.420\inference_engine\samples\Samples.sln
- 大小: 51 KB
- 大小: 56.6 KB
- 大小: 95.9 KB
- 大小: 105.9 KB
- 大小: 187.3 KB
分享到:
相关推荐
Intel Movidius Neural Compute Stick 跑tensorflow训练的mnist模型的例子,博客链接: https://blog.csdn.net/rookie_wei/article/details/98873920
RBF+Neural+Network+Control+for+Mechanical+Systems-Matlab+programs
Neural+Networks+and+Deep+Learning-神经网络与深度学习.pdf
The Detailed Data on the Neural Compute Stick Acceleration Performance
Coursera Machine Learning 第五周week5 ex4Neural Networks Learning编程全套满分题目+注释
原英文网站:...中文翻译:https://www.gitbook.com/book/hit-scir/neural-networks-and-deep-learning-zh_cn/details 方便无梯子的同学:链接: https://pan.baidu.com/s/12UVkF4M1WUPAvZC28BH0vA 密码: hq7x
+ Artificial neural network ( ANN ) + Feed Forward Neural Network ( FFNN ) + Cascade Forward Neural Network ( CFNN ) + Recurrent Neural Network ( RNN ) + Generalized Regression Neural Network ( ...
neural network design 2nd edtion Copyright by Martin T. Hagan and Howard B. Demuth. All rights reserved. No part of the book may be reproduced, stored in a retrieval system, or transcribed in any form...
Neuraltalk2-pytorch:在pytorch中的图像字幕模型
在openVino深度学习框架下,利用Intel Neural Compute Stick神经计算棒,实现接近实时的物体检测,约5PFS(根据不同的硬件设备)
ISSCC2021_Session_15V_Compute-in-Memory Processors for Deep Neural Networks.pdf
C++ Neural Networks And Fuzzy Logic 2nd by Valluru Rao(英文pdf+源码) C++ Neural Networks And Fuzzy Logic 2nd by Valluru Rao (source code for DOS) MIS press 1995 这本书有约略提到类神经网络如何应用在...
A step-by-step gentle journey through the mathematics of neural networks, and making your own using the Python computer language. Neural networks are a key element of deep learning and artificial ...
Neural-Net:具有用C ++制成的多层隐藏层的神经网络
Neural Enhancement Suite是一款基于人工智能AI,可以在AE中对视频进行降噪、视频上色、锐化、弱光亮度提升等操作 Neural Enhancement Suite is an A.I. powered toolset for AI-based video enhancement to ...
NeuralNet_publish2014-10-22 自己动手写神经网络
完成训练后,可以使用一台计算机(Linux,macOS或Windows)作为TensorRider的计算服务器,令实验小车在线运行,也可以通过在TensorRider的Raspberry Pi上连接Intel Movidius Neural Compute Stick(NCS),实现离线...
总共1000多页,很好的资料,着重讲DL4J。