Skip to content

yao-matrix/caffe_

Repository files navigation

Add-on Notes

This branch of Caffe extend Microsoft Caffe by integrating MKL DNN API accelerations and doing many extra OpenMP/Zero-Copy optimizations to make py-RFCN fasterin Intel platform. We also implemented the CPU version of some layers(like PSROI, box annotator etc.). We can get 20x acceleration compared with Vanilla CPU Caffe in Pascal-VOC RFCN end2end case in Xeon E5 2699-v4.

How to use MKL's DNN API

  1. go to ./external/mkl and run prepare_mkl.sh. Once it's done, you can see a mklml_lnx_2017.0.2.20161122 folder in ./external/mkl directory
  2. Update BLAS_INCLUDE and BLAS_LIB's absolute path part in Makefile.config

Build

  1. make -j8
  2. make pycaffe

This has been tested in CentOS 7.2, suppose no issues in Linux based OS.

Contact: Matrix YAO([email protected])


Caffe

Linux (CPU) Windows (CPU)
Travis Build Status AppVeyor Build Status

License

Caffe is a deep learning framework made with expression, speed, and modularity in mind. It is developed by the Berkeley Vision and Learning Center (BVLC) and community contributors.

Check out the project site for all the details like

and step-by-step examples.

License and Citation

Caffe is released under the BSD 2-Clause license. The BVLC reference models are released for unrestricted use.

Please cite Caffe in your publications if it helps your research:

@article{jia2014caffe,
  Author = {Jia, Yangqing and Shelhamer, Evan and Donahue, Jeff and Karayev, Sergey and Long, Jonathan and Girshick, Ross and Guadarrama, Sergio and Darrell, Trevor},
  Journal = {arXiv preprint arXiv:1408.5093},
  Title = {Caffe: Convolutional Architecture for Fast Feature Embedding},
  Year = {2014}
}