Skip to content
Evan Shelhamer edited this page Apr 14, 2017 · 7 revisions

Projects Using Caffe

Fast R-CNN is a fast framework for object detection with deep ConvNets. Fast R-CNN

  • trains state-of-the-art models, like VGG16, 9x faster than traditional R-CNN and 3x faster than SPPnet,
  • runs 200x faster than R-CNN and 10x faster than SPPnet at test-time,
  • has a significantly higher mAP on PASCAL VOC than both R-CNN and SPPnet,
  • and is written in Python and C++/Caffe.

Please take a look at @rbgirshick work at https://github.com/rbgirshick/fast-rcnn


The Facebook Caffe Extensions include the predictor inference API, a torch2caffe translator, and other conversion scripts and tools. Thanks to @ajtulloch.


The NVIDIA GPU Rest Engine demonstrates a server for low-latency image classification inference. It is a technical demo that shows how you can add a REST API on top of Caffe using the Go language, and how to package all your dependencies inside a Docker container. Thanks to @flx42.


Ristretto is a framework for deep network approximation that can help experiment with quantization, reduced precision, and other optimizations.



Expresso screenshot Expresso is a Python-based GUI for designing, training and using CNNs. Expresso uses Caffe as its backend CNN framework. Some of its salient features :

  • A convenient wizard-like interface to contextually guide the user during common scenarios such as data import, design and training of CNNs
  • A smart-edit interface makes net creation easy and quick.
  • Deep networks are color-coded and informatively presented
  • Support for training external classifier (SVM) using deep features (i.e. features extracted by passing image data through pre-trained CNN and tapping output at layer(s) of CNN).
  • Data Visualization

Visit the project page for installation details and links to text/video tutorials.


Deep Visualization Toolbox is an open source software tool that lets you probe deep networks by feeding them an image (or a live webcam feed) and watching the reaction of every unit. You can also select individual units to view pre-rendered visualizations of what that neuron “wants to see most”.

Visit the deep-vis project page and the @yosinski project at https://github.com/yosinski/deep-visualization-toolbox


Caffe with Spearmint integrates the deep learning framework with spearmint bayesian hyperparamter optimization to automate parameter search.

Visit the @kuz project at https://github.com/kuz/caffe-with-spearmint

Other Useful Components

Clone this wiki locally