https://github.com/MartinHjelm/fishercaffe

C++ training and testing code for an SVM using Vlfeat fisher vectors together with possible other features.
https://github.com/MartinHjelm/fishercaffe

Last synced: 3 months ago
JSON representation

C++ training and testing code for an SVM using Vlfeat fisher vectors together with possible other features.

• Host: GitHub
• URL: https://github.com/MartinHjelm/fishercaffe
• Owner: MartinHjelm
• Created: 2016-05-22T16:06:00.000Z (about 8 years ago)
• Default Branch: master
• Last Pushed: 2016-06-29T23:22:59.000Z (almost 8 years ago)
• Last Synced: 2024-03-19T23:20:31.302Z (3 months ago)
• Language: C++
• Size: 10.7 KB
• Stars: 14
• Watchers: 3
• Forks: 1
• Open Issues: 1
• Metadata Files:
  • Readme: README.md

Lists

• my-awesome-awesomeness - fishercaffe

README

        
# Fisher Caffe

C++ implementation of learning Fisher vectors using VlFeat. In addition the fisher vector can be concatenated with another feature vector such as a Caffe layer etc. 

## Requirements

* CMake

* Boost

* Eigen

* OpenCV

* VlFeat

## Training

* Extract Dense SIFTs from a set of labeled images

* PCA Project

* Compute GMM using EM

* Compute Fisher Vector for the images from the trained GMM

* Add additional features.

* Train one-against-many linear SVM for each class

## Test

* Extract Dense SIFTs from the set of test images

* PCA Project

* Compute Fisher Vector for the images from the trained GMM

* Add additional features.

* Classify using linear SVM

This code is provided as is. If you notice any errors or bugs file a request or inbox me.