Single-shot cuboids: Geodesics-based end-to-end Manhattan aligned layout estimation from spherical panoramas

作者：

Highlights：

• Single-shot, end-to-end layout corner estimation model from spherical panoramas.

• Direct keypoint estimation allows for the integration of explicit layout constraints.

• Geodesic distance loss and Gaussian heatmaps for spherical center of mass keypoints.

• Homography-based module that ensures end-to-end full Manhattan alignment.

摘要

•Single-shot, end-to-end layout corner estimation model from spherical panoramas.•Direct keypoint estimation allows for the integration of explicit layout constraints.•Geodesic distance loss and Gaussian heatmaps for spherical center of mass keypoints.•Homography-based module that ensures end-to-end full Manhattan alignment.

论文关键词：Panoramic scene understanding,Indoor 3D reconstruction,Layout estimation,Spherical panoramas,Omnidirectional vision

论文评审过程：Received 10 August 2020, Revised 20 January 2021, Accepted 11 March 2021, Available online 18 March 2021, Version of Record 12 April 2021.

论文官网地址：https://doi.org/10.1016/j.imavis.2021.104160