Skip to content

This repository was a fork of BVLC/caffe and includes the upsample, bn, dense_image_data and softmax_with_loss (with class weighting) layers of caffe-segnet (https://github.com/alexgkendall/caffe-segnet) to run SegNet with cuDNN version 5.

License

Notifications You must be signed in to change notification settings

rockcdr/caffe-segnet-cudnn5

 
 

Repository files navigation

Caffe SegNet cuDNN5

This is a modified version of Caffe which supports the SegNet architecture

As described in SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation Vijay Badrinarayanan, Alex Kendall and Roberto Cipolla [http://arxiv.org/abs/1511.00561]

Please refer to Alex Kendalls caffe-segnet for tutorial and a guide how to use it (https://github.com/alexgkendall/caffe-segnet).

Since the original caffe-segnet supports just cuDNN v2, which is not supported for new pascal based GPUs, it was possible to decrease the inference time by 25 % to 35 % with caffe-segnet-cudnn5 using Titan X Pascal.

I recommend to use my trained weights (CityScapes Model) for semantic segmenation of traffic scenes, which you can find in segnet model zoo: https://github.com/alexgkendall/SegNet-Tutorial/blob/master/Example_Models/segnet_model_zoo.md

If you like to use SegNet with C++, the test_segmentation.cpp might be helpful. https://github.com/alexgkendall/SegNet-Tutorial/blob/master/Scripts/test_segmentation.cpp

News

cuDNN v.6 has been released. I have tested it using Titan X Pascal. It doesn't bring any noticeable improvements for SegNet. For that reason I will not update the repository to cuDNN6.

Publications

If you use this software in your research, please cite their publications:

http://arxiv.org/abs/1511.02680 Alex Kendall, Vijay Badrinarayanan and Roberto Cipolla "Bayesian SegNet: Model Uncertainty in Deep Convolutional Encoder-Decoder Architectures for Scene Understanding." arXiv preprint arXiv:1511.02680, 2015.

http://arxiv.org/abs/1511.00561 Vijay Badrinarayanan, Alex Kendall and Roberto Cipolla "SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation." arXiv preprint arXiv:1511.00561, 2015.

License

This extension to the Caffe library is released under a creative commons license which allows for personal and research use only. For a commercial license please contact the authors. You can view a license summary here: http://creativecommons.org/licenses/by-nc/4.0/

About

This repository was a fork of BVLC/caffe and includes the upsample, bn, dense_image_data and softmax_with_loss (with class weighting) layers of caffe-segnet (https://github.com/alexgkendall/caffe-segnet) to run SegNet with cuDNN version 5.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • C++ 79.6%
  • Python 8.1%
  • Cuda 6.0%
  • CMake 2.7%
  • Protocol Buffer 1.7%
  • MATLAB 0.9%
  • Other 1.0%