Gaze latent support vector machine for image classification improved by weakly supervised region selection

作者：

Highlights：

• A novel WSL model integrating human gaze annotation for image classification is proposed.

• Our model generalizes positive instance Gaze latent SVM.

• Our model generalizes top K instance latent SVM.

• Our model is validated on extensive experiments in various challenging contexts.

• A new food-related dataset with gaze annotation is published.

摘要

•A novel WSL model integrating human gaze annotation for image classification is proposed.•Our model generalizes positive instance Gaze latent SVM.•Our model generalizes top K instance latent SVM.•Our model is validated on extensive experiments in various challenging contexts.•A new food-related dataset with gaze annotation is published.

论文关键词：Weakly supervised learning,Human gaze,image classification

论文评审过程：Received 10 February 2017, Revised 27 June 2017, Accepted 1 July 2017, Available online 3 July 2017, Version of Record 7 July 2017.

论文官网地址：https://doi.org/10.1016/j.patcog.2017.07.001