Caffe: Convolutional Architecture for Fast Feature Embedding

The paper presents Caffe, a clean and modifiable framework for state-of-the-art deep learning algorithms together with a collection of reference models. Caffe is a BSD-licensed C++ library with Python and MATLAB bindings for training and deploying general-purpose convolutional neural networks and other deep models efficiently on commodity architectures. Its core design separates model representation from the actual implementation, allowing experimentation and seamless switching among platforms for ease of development and deployment, from prototyping machines to cloud environments.

Through CUDA GPU computation, Caffe fits industry and internet-scale media needs, processing over 40 million images a day on a single K40 or Titan GPU, approximately 2 ms per image. Maintained and developed by the Berkeley Vision and Learning Center with the help of an active community of contributors on GitHub, it powers ongoing research projects, large-scale industrial applications, and startup prototypes in vision, speech, and multimedia.