Learning spatial relations and shapes for structural object description and scene recognition

作者：

Highlights：

• We propose a graph-based multilevel descriptor for object recognition.

• This descriptor allows to characterize spatial relations and shapes between objects.

• An original bags-of-features strategy to learn such descriptors is proposed.

• Experiments on four datasets of images highlight the interest of the approach.

摘要

•We propose a graph-based multilevel descriptor for object recognition.•This descriptor allows to characterize spatial relations and shapes between objects.•An original bags-of-features strategy to learn such descriptors is proposed.•Experiments on four datasets of images highlight the interest of the approach.

论文关键词：Spatial relations,Relative position descriptors,Bags of relations,Structural object description,Hierarchical representation,Force histograms

论文评审过程：Received 25 April 2017, Revised 17 April 2018, Accepted 25 June 2018, Available online 20 July 2018, Version of Record 20 July 2018.

论文官网地址：https://doi.org/10.1016/j.patcog.2018.06.017