Multi-modal deep feature learning for RGB-D object detection

作者：

Highlights：

• We present an approach for RGB-D object detection, which can exploit both modality-correlated and modality-specific relationships between RGB and depth images.

• The shared weights strategy and a parameter-free-correlation layer are introduced to extract the modality-correlated representations.

• The proposed approach can simultaneously generate RGB-D region proposals and perform region-wise RGB-D object recognition.

摘要

•We present an approach for RGB-D object detection, which can exploit both modality-correlated and modality-specific relationships between RGB and depth images.•The shared weights strategy and a parameter-free-correlation layer are introduced to extract the modality-correlated representations.•The proposed approach can simultaneously generate RGB-D region proposals and perform region-wise RGB-D object recognition.

论文关键词：RGB-D objectness estimation,RGB-D object detection,Multi-modal learning,Convolutional neural networks

论文评审过程：Received 29 December 2016, Revised 10 June 2017, Accepted 26 July 2017, Available online 29 July 2017, Version of Record 17 August 2017.

论文官网地址：https://doi.org/10.1016/j.patcog.2017.07.026