Three-dimensional (3D) computer vision is a fundamental backbone technology of autonomous vehicles, robots and the emerging reality technologies for interaction and presence. Spherical cameras are holistic images from a viewpoint perspective and as such, very well posed to perceive the world geometrically. Modern computer vision is driven by data and has managed to overcome barriers that were seemingly impossible, one of which is monocular geometric inference. This thesis focuses on monocular geometric inference from a single spherical panorama, and particularly, 3D reconstructing indoor spaces. It applies spherical geometry and domain knowledge to improve the performance of modern data-driven technology. To achieve this, parallel contr...
Spherical cameras capture scenes in a holistic manner and have been used for room layout estimation....
Modern 3D vision advancements rely on data driven methods and thus, task specific annotated datasets...
Spherical cameras capture all the information around them with a field of view (FOV) of 360° 180°. T...
Learning based approaches for depth perception are limited by the availability of clean training dat...
Recent work on depth estimation up to now has only focused on projective images ignoring 360o conten...
Image-based three-dimensional (3D) scene reconstruction approaches have been widely studied by the s...
Omnidirectional vision is becoming increasingly relevant as more efficient 360° image acquisition is...
A dataset of omnidirectional (360 - spherical panoramas) images with their corresponding ground tru...
International audienceThis paper presents a method and apparatus for building dense visual maps of l...
In this paper, we propose to work in the 2.5D space of the scene to facilitate composition of new sp...
Modern 3D vision advancements rely on data driven methods and thus, task specific annotated datasets...
International audienceThe current work addresses the problem of 3D model tracking in the context of ...
A estimativa de profundidade é um componente essencial de diversas aplicações de visão computacional...
We propose a 3D environment modelling method using multiple pairs of high-resolution spherical image...
We propose a 3D environment modelling method using multiple pairs of high-resolution spherical image...
Spherical cameras capture scenes in a holistic manner and have been used for room layout estimation....
Modern 3D vision advancements rely on data driven methods and thus, task specific annotated datasets...
Spherical cameras capture all the information around them with a field of view (FOV) of 360° 180°. T...
Learning based approaches for depth perception are limited by the availability of clean training dat...
Recent work on depth estimation up to now has only focused on projective images ignoring 360o conten...
Image-based three-dimensional (3D) scene reconstruction approaches have been widely studied by the s...
Omnidirectional vision is becoming increasingly relevant as more efficient 360° image acquisition is...
A dataset of omnidirectional (360 - spherical panoramas) images with their corresponding ground tru...
International audienceThis paper presents a method and apparatus for building dense visual maps of l...
In this paper, we propose to work in the 2.5D space of the scene to facilitate composition of new sp...
Modern 3D vision advancements rely on data driven methods and thus, task specific annotated datasets...
International audienceThe current work addresses the problem of 3D model tracking in the context of ...
A estimativa de profundidade é um componente essencial de diversas aplicações de visão computacional...
We propose a 3D environment modelling method using multiple pairs of high-resolution spherical image...
We propose a 3D environment modelling method using multiple pairs of high-resolution spherical image...
Spherical cameras capture scenes in a holistic manner and have been used for room layout estimation....
Modern 3D vision advancements rely on data driven methods and thus, task specific annotated datasets...
Spherical cameras capture all the information around them with a field of view (FOV) of 360° 180°. T...