Quantifying the Uncertainty for Neural Radiance Fields in Few-Shot Scenarios

Labreche, M.T. (2023) Quantifying the Uncertainty for Neural Radiance Fields in Few-Shot Scenarios. Bachelor's Thesis, Artificial Intelligence.

Preview	Text bAI_2023_LabrecheMT.pdf Download (8MB) | Preview
	Text toestemming.pdf Restricted to Registered users only Download (126kB)

Abstract

Neural Radiance Fields (NeRFs) can accurately capture 3D volumes in neural net works but are often unreliable in real-world settings where information is limited. This is prob-lematic for settings that require decision-making like autonomous driving. This research uses uncertainty quantification to gain insight into the reliability of predictions. This is done by applying a method called Flipout to NeRF. It uses Variational Inference to increase uncertainty quality and is found to produce better predictions and associated uncertainty than previous methods.

Item Type:	Thesis (Bachelor's Thesis)
Supervisor name:	Valdenegro Toro, M.A.
Degree programme:	Artificial Intelligence
Thesis type:	Bachelor's Thesis
Language:	English
Date Deposited:	16 Aug 2023 09:13
Last Modified:	16 Aug 2023 09:13
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/31183

Actions (login required)

View Item