Depth self-supervision for single image novel view synthesis
Not on the map. Why?
Abstract (arXiv)
In this paper, we tackle the problem of generating a novel image from an arbitrary viewpoint given a single frame as input. While existing methods operating in this setup aim at predicting the target view depth map to guide the synthesis, without explicit supervision over such a task, we jointly optimize our framework for both novel view synthesis and depth estimation to unleash the synergy between the two at its best. Specifically, a shared depth decoder is trained in a self-supervised manner to predict depth maps that are consistent across the source and target views. Our results demonstrate the effectiveness of our approach in addressing the challenges of both tasks allowing for higher-quality generated images, as well as more accurate depth for the target viewpoint.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 37 of its 60 references that are in the map. First posted on arXiv 27 Aug 2023.
Builds on
37 of 60 references.
References · 37
- 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,713
- 2018The Unreasonable Effectiveness of Deep Features as a Perceptual Metriccited by 11,177
- 2022Instant Neural Graphics Primitives with a Multiresolution Hash Encodingcited by 3,723
- 2015ShapeNet: An Information-Rich 3D Model Repositorycited by 2,897
- 2021Plenoxels: Radiance Fields without Neural Networkscited by 1,613
- 2020pixelNeRF: Neural Radiance Fields from One or Few Imagescited by 1,584
- 2015Spatial Transformer Networkscited by 1,429
- 2014Depth Map Prediction from a Single Image using a Multi-Scale Deep Networkcited by 1,274
- 2021Direct Voxel Grid Optimization: Super-fast Convergence for Radiance Fields Reconstructioncited by 964
- 2016Unsupervised Monocular Depth Estimation with Left-Right Consistencycited by 820
- 2021MVSNeRF: Fast Generalizable Radiance Field Reconstruction from Multi-View Stereocited by 740
- 2021IBRNet: Learning Multi-View Image-Based Renderingcited by 703
25 more
- 20163D-R2N2: A Unified Approach for Single and Multi-view 3D Object Reconstructioncited by 588
- 2018DeepVoxels: Learning Persistent 3D Feature Embeddingscited by 405
- 2019SynSin: End-to-end View Synthesis from a Single Imagecited by 364
- 2019Neural Point-Based Graphicscited by 327
- 2020Free View Synthesiscited by 231
- 2020GRF: Learning a General Radiance Field for 3D Representation and Renderingcited by 183
- 2020Stable View Synthesiscited by 165
- 20193D Ken Burns Effect from a Single Imagecited by 128
- 2019Neural Rerendering in the Wildcited by 121
- 2018Extreme View Synthesiscited by 94
- 2021PixelSynth: Generating a 3D-Consistent Experience from a Single Imagecited by 82
- 2020Worldsheet: Wrapping the World in a 3D Sheet for View Synthesis from a Single Imagecited by 77
- 2017Learning to Synthesize a 4D RGBD Light Field from a Single Imagecited by 60
- 2017Transformation-Grounded Image Generation Network for Novel 3D View Synthesiscited by 59
- 2017Factoring Shape, Pose, and Layout from the 2D Image of a 3D Scenecited by 50
- 2022FWD: Real-time Novel View Synthesis with Forward Warping and Depthcited by 39
- 2021Fast and Explicit Neural View Synthesiscited by 28
- 2018Geometry-aware Deep Network for Single-Image Novel View Synthesiscited by 19
- 2021Self-Supervised Visibility Learning for Novel View Synthesiscited by 14
- 2016Novel Views of Objects from a Single Imagecited by 12
- 2022Neural Mesh-Based Graphicscited by 12
- 2018View Extrapolation of Human Body from a Single Imagecited by 11
- 2021NVS-MonoDepth: Improving Monocular Depth Prediction with Novel View Synthesiscited by 7
- 2021Novel View Synthesis via Depth-guided Skip Connectionscited by 5
- 2022Novel View Synthesis for High-fidelity Headshot Scenescited by 4
Built on by
Not cited yet.