A contextual conditional random field network for monocular depth estimation

作者：

Highlights：

• Deep convolutional neural network for monocular depth estimation.

• Skip connection enhance the details of output, but also introduces some noises to estimated depth map.

• Conditional random field are adopted to alleviate these noises by regularizing the information flow from the encoder to the decoder.

• Long-tail distribution in depth map is alleviated by the depth-guided weights.

摘要

•Deep convolutional neural network for monocular depth estimation.•Skip connection enhance the details of output, but also introduces some noises to estimated depth map.•Conditional random field are adopted to alleviate these noises by regularizing the information flow from the encoder to the decoder.•Long-tail distribution in depth map is alleviated by the depth-guided weights.

论文关键词：Monocular depth estimation,Deep neural network,Skip connection,Conditional random field

论文评审过程：Received 16 September 2019, Revised 4 January 2020, Accepted 28 March 2020, Available online 23 April 2020, Version of Record 29 April 2020.

论文官网地址：https://doi.org/10.1016/j.imavis.2020.103922