Tampere University of Technology

TUTCRIS Research Portal

Predicting Novel Views Using Generative Adversarial Query Network

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

Details

Original languageEnglish
Title of host publicationImage Analysis - 21st Scandinavian Conference, SCIA 2019, Proceedings
EditorsMichael Felsberg, Per-Erik Forssén, Jonas Unger, Ida-Maria Sintorn
PublisherSpringer Verlag
Pages16-27
Number of pages12
ISBN (Print)9783030202040
DOIs
Publication statusPublished - 2019
Publication typeA4 Article in a conference publication
EventScandinavian Conference on Image Analysis - Norrköping, Sweden
Duration: 11 Jun 201913 Jun 2019

Publication series

NameLecture Notes in Computer Science
Volume11482
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

ConferenceScandinavian Conference on Image Analysis
CountrySweden
CityNorrköping
Period11/06/1913/06/19

Abstract

The problem of predicting a novel view of the scene using an arbitrary number of observations is a challenging problem for computers as well as for humans. This paper introduces the Generative Adversarial Query Network (GAQN), a general learning framework for novel view synthesis that combines Generative Query Network (GQN) and Generative Adversarial Networks (GANs). The conventional GQN encodes input views into a latent representation that is used to generate a new view through a recurrent variational decoder. The proposed GAQN builds on this work by adding two novel aspects: First, we extend the current GQN architecture with an adversarial loss function for improving the visual quality and convergence speed. Second, we introduce a feature-matching loss function for stabilizing the training procedure. The experiments demonstrate that GAQN is able to produce high-quality results and faster convergence compared to the conventional approach.

Keywords

  • Generative Adversarial Query Network, Mean feature matching loss, Novel view synthesis

Publication forum classification

Field of science, Statistics Finland