Towards effective codebookless model for image classification

作者：

Highlights：

• We show that codebookless model (CLM) is a very competitive alternative to the mainstream BoF model.

• Two well-motivated parameters are introduced to further improve our CLM.

• A joint low-rank learning and SVM is proposed on Gaussian manifold.

• The comprehensive experiments demonstrated the promising performance of our CLM.

摘要

Highlights•We show that codebookless model (CLM) is a very competitive alternative to the mainstream BoF model.•Two well-motivated parameters are introduced to further improve our CLM.•A joint low-rank learning and SVM is proposed on Gaussian manifold.•The comprehensive experiments demonstrated the promising performance of our CLM.

论文关键词：Codebookless model,Image classification,Bag-of-features,Riemannian manifold

论文评审过程：Received 23 July 2015, Revised 28 February 2016, Accepted 2 March 2016, Available online 11 March 2016, Version of Record 23 August 2016.

论文官网地址：https://doi.org/10.1016/j.patcog.2016.03.004