FreDSNet: Joint Monocular Depth and Semantic Segmentation with Fast Fourier Convolutions
arxiv(2022)
摘要
In this work we present FreDSNet, a deep learning solution which obtains
semantic 3D understanding of indoor environments from single panoramas.
Omnidirectional images reveal task-specific advantages when addressing scene
understanding problems due to the 360-degree contextual information about the
entire environment they provide. However, the inherent characteristics of the
omnidirectional images add additional problems to obtain an accurate detection
and segmentation of objects or a good depth estimation. To overcome these
problems, we exploit convolutions in the frequential domain obtaining a wider
receptive field in each convolutional layer. These convolutions allow to
leverage the whole context information from omnidirectional images. FreDSNet is
the first network that jointly provides monocular depth estimation and semantic
segmentation from a single panoramic image exploiting fast Fourier
convolutions. Our experiments show that FreDSNet has similar performance as
specific state of the art methods for semantic segmentation and depth
estimation. FreDSNet code is publicly available in
https://github.com/Sbrunoberenguel/FreDSNet
更多查看译文
关键词
360-degree contextual information,additional problems,addressing scene understanding problems,context information,convolutional layer,deep learning solution,entire environment,fast Fourier convolutions,FreDSNet code,good depth estimation,indoor environments,joint monocular depth,monocular depth estimation,omnidirectional images,semantic 3D understanding,semantic segmentation,single panoramas,single panoramic image,task-specific advantages,wider receptive field
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要