DL架構之darknet：深度學習架構darknet的簡介、安裝、使用方法的詳細攻略

目錄

darknet的簡介

darknet的安裝

darknet的使用方法

１、YOLO: Real-Time Object Detection

2、ImageNet Classification

3、Nightmare

4、RNNs in Darknet

5、DarkGo: Go in Darknet

6、Train a Classifier on CIFAR-10

7、Hardware Guide: Neural Networks on GPUs (Updated 2016-1-30)

      Darknet is an open source neural network framework written in C and CUDA. It is fast, easy to install, and supports CPU and GPU computation.

For more information see the Darknet project website.

For questions or issues please use the Google Group.

       Darknet: Open Source Neural Networks in C.  Darknet is an open source neural network framework written in C and CUDA. It is fast, easy to install, and supports CPU and GPU computation. You can find the source on GitHub or you can read more about what Darknet can do right here.

官網：

https://pjreddie.com/darknet/

GitHub：

https://github.com/pjreddie/darknet

      darknet是一個較為輕型的完全基于C與CUDA的開源深度學習架構，其主要特點就是容易安裝，沒有任何依賴項（OpenCV都可以不用），移植性非常好，支援CPU與GPU兩種計算方式。

      相比于TensorFlow來說，darknet并沒有那麼強大，但這也成了darknet的優勢：  

darknet完全由C語言實作，沒有任何依賴項，當然可以使用OpenCV，但隻是用其來顯示圖檔、為了更好的可視化；  

darknet支援CPU（是以沒有GPU也不用緊的）與GPU（CUDA/cuDNN，使用GPU當然更塊更好了）；  

       正是因為其較為輕型，沒有像TensorFlow那般強大的API，是以給我的感覺就是有另一種味道的靈活性，适合用來研究底層，可以更為友善的從底層對其進行改進與擴充；  darknet的實作與caffe的實作存在相似的地方，熟悉了darknet，相信對上手caffe有幫助；

git clone

後期繼續更新……

@misc{darknet13,

 author =   {Joseph Redmon},

 title =    {Darknet: Open Source Neural Networks in C},

 howpublished = {\url{http://pjreddie.com/darknet/}},

 year = {2013--2016}

}

You only look once (YOLO) is a state-of-the-art, real-time object detection system.

Classify images with popular models like ResNet and ResNeXt.

Use Darknet's black magic to conjure ghosts, ghouls, and wild badgermoles. But be warned, ye who enter here: no one is safe in the land of nightmares.

Recurrent neural networks are all the rage for time-series data and NLP. Learn how to use them in Darknet!

Play Go using a policy network trained with Darknet

Learn how to train a classifier from scratch in Darknet.

I've had a number of people ask me what hardware I would recommend for training neural networks for vision applications. Here are some of my thoughts.