A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
Filters
Weakly-supervised Single-view Image Relighting
[article]
2023
arXiv
pre-print
Preserved Fulltext
We present a learning-based approach to relight a single image of Lambertian and low-frequency specular objects. ...
To facilitate the weakly-supervised training, we contribute Relit, a large-scale (750K images) dataset of videos with aligned objects under changing illuminations. ...
We demonstrate single-object insertion and multi-object insertion where multiple objects are from different input images. We also demonstrate editing the materials of objects. ...
arXiv:2303.13852v1
fatcat:yzxl52mkong4pgwrgq5glui6ty
Joint Learning of Portrait Intrinsic Decomposition and Relighting
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
To solve this ill-posed problem from single image, state-of-the-art methods in shape from shading mostly resort to supervised training on all the components on either synthetic or real datasets. ...
Here, we propose a new self-supervised training paradigm that 1) reduces the need for full supervision on the decomposition task and 2) takes into account the relighting task. ...
Method Our goal is to decompose a single-view RGB image I into its intrinsic components including albedo A, normal N and lighting L. ...
arXiv:2106.15305v1
fatcat:27ezbp2lbrgb7b6i5iwrr54wua
Learning Inverse Rendering of Faces from Real-world Videos
[article]
2020
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Meanwhile, since no ground truth for any component is available for real images, it is not feasible to conduct supervised learning on real face images. ...
To alleviate this problem, we propose a weakly supervised training approach to train our model on real face videos, based on the assumption of consistency of albedo and normal across different frames, ...
(Best viewed in PDF with zoom.) Fig. 8: The left of the figure is case of photo relighting. ...
arXiv:2003.12047v1
fatcat:g2oeps3csvf4niqucfxjxwrpwu
DeepPS2: Revisiting Photometric Stereo Using Two Differently Illuminated Images
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Other Versions
We demonstrate how image relighting in conjunction with image reconstruction enhances the lighting estimation in a self-supervised setting. ...
We propose an inverse rendering-based deep learning framework, called DeepPS2, that jointly performs surface normal, albedo, lighting estimation, and image relighting in a completely self-supervised manner ...
While LERPS [42] infers lighting and surface normal from a single image, it requires multiple images (one at a time) for training. ...
arXiv:2207.02025v2
fatcat:kcbujlxvq5dyjc45v6hjx3ab6a
Structured 3D Features for Reconstructing Controllable Avatars
[article]
2023
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
Other Versions
Moreover, we show that the proposed methodology allows novel view synthesis, relighting, and re-posing the reconstruction, and can naturally be extended to handle multiple input images (e.g. different ...
illumination decomposition, as a result of a single end-to-end model, trained semi-supervised, and with no additional postprocessing. ...
The image used as input in the single-view case is shown in the second column. ...
arXiv:2212.06820v2
fatcat:tym7y2bvmjg2jfgjkydkua3uka
De-rendering 3D Objects in the Wild
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Other Versions
We present a weakly supervised method that is able to decompose a single image of an object into shape (depth and normals), material (albedo, reflectivity and shininess) and global lighting parameters. ...
This shape supervision can come for example from a pretrained depth network or - more generically - from a traditional structure-from-motion pipeline. ...
Single Image Relighting. To demonstrate the usefulness of de-rendering, we perform relighting on the CelebA-HQ dataset. ...
arXiv:2201.02279v2
fatcat:u5gcbslfrrh6pjvzk3ppuxw7dm
Physics-based Shadow Image Decomposition for Shadow Removal
[article]
2020
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Note that this fulltext copy is not of the "primary" version of this work. The version it corresponds to is:
2021 accepted arXiv:2012.13018v2 fatcat:uvtlusga3jgipgsktgl22s6et4
Furthermore, this decomposition allows us to formulate a patch-based weakly-supervised shadow removal method. ...
paired shadow and shadow-free images. ...
Weakly-supervised Shadow Removal Evaluation We compare our weakly-supervised approach with earlier priorbased and weakly-supervised shadow removal methods in Tab. 2. The methods of Yang et al. ...
arXiv:2012.13018v1
fatcat:4xlscbzxknhflcu3bmamckx4pq
GAN2X: Non-Lambertian Inverse Rendering of Image GANs
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Other Versions
single-view 3D face reconstruction. ...
Recovering these underlying intrinsic components from 2D images, also known as inverse rendering, usually requires a supervised setting with paired images collected from multiple viewpoints and lighting ...
We thank Abhimitra Meka for providing the Total Relighting results. ...
arXiv:2206.09244v4
fatcat:3fis2ihwb5g3tkhzy76gphzdgi
I^2-SDF: Intrinsic Indoor Scene Reconstruction and Editing via Raytracing in Neural SDFs
[article]
2023
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2023; you can also visit the original URL.
The file type is application/pdf.
Other Versions
Our holistic neural SDF-based framework jointly recovers the underlying shapes, incident radiance and materials from multi-view images. ...
spatially-varying material of the scene as a neural field through surface-based, differentiable Monte Carlo raytracing and emitter semantic segmentations, which enables physically based and photorealistic scene relighting ...
Since the ground truths of material are impossible to capture from images, we weakly supervise the network by re-render results instead of direct supervision from strong material priors. ...
arXiv:2303.07634v2
fatcat:yedqis5ykvbnxca7ks3gmtaoa4
Towards High Fidelity Monocular Face Reconstruction with Rich Reflectance using Self-supervised Learning and Ray Tracing
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Note that this fulltext copy is not of the "primary" version of this work. The version it corresponds to is:
2021 pre-print arXiv:2103.15432v2 fatcat:h53iokeyufdy7kp477onmgchieOther Versions
They can also be trained in self-supervised manner for increased robustness and better generalization. ...
Robust face reconstruction from monocular image in general lighting conditions is challenging. ...
Disentangling light color from skin color from a single image is an ill-posed problem and is not solved in this work. ...
arXiv:2103.15432v3
fatcat:g2c7x3ellreztprlmlm6e6bz3m
Advances in Neural Rendering
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Other Versions
Neural rendering is a leap forward towards the goal of synthesizing photo-realistic image and video content. ...
Synthesizing photo-realistic images and videos is at the heart of computer graphics and has been the focus of decades of research. ...
It assumes a pre-defined set of multi-view images given, though, at training time, only a single view chosen arbitrarily is used for supervision at any time. ...
arXiv:2111.05849v2
fatcat:nbvkfg2bjvgqdopdqwl33rt4ii
DisUnknown: Distilling Unknown Factors for Disentanglement Learning
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
However, it is often expensive or even impossible to label every single factor to achieve fully-supervised disentanglement. ...
Under this setting, we propose a flexible weakly-supervised multi-factor disentanglement framework DisUnknown, which Distills Unknown factors for enabling multi-conditional generation regarding both labeled ...
The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the Army Research Office ...
arXiv:2109.08090v1
fatcat:pxi5d22amfhjlakvwj7x4irqma
Author Index
2010
2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2017; you can also visit the original URL.
The file type is application/pdf.
Weakly Supervised Learning from Multiple Imperfect Oracles
Liu, Shuaicheng
Super Resolution using Edge Prior and Single Image Detail Synthesis
Liu, Shubao
Ray Markov Random Fields for Image-Based ...
Learning for Image Classification Improving Web Image Search Results using Query-relative Classifiers Vezhnevets, Alexander Towards Weakly Supervised Semantic Segmentation by Means of Multiple Instance ...
doi:10.1109/cvpr.2010.5539913
fatcat:y6m5knstrzfyfin6jzusc42p54
DeRenderNet: Intrinsic Image Decomposition of Urban Scenes with Shape-(In)dependent Shading Rendering
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
We propose DeRenderNet, a deep neural network to decompose the albedo and latent lighting, and render shape-(in)dependent shadings, given a single image of an outdoor urban scene, trained in a self-supervised ...
supervision. ...
Recently, a complex self-supervised outdoor scene decomposition framework [14] has been proposed to further separate shadows from the shading image with a multi-view training dataset, and relighting ...
arXiv:2104.13602v1
fatcat:ocr63ed7wrdundkruylvt7uyia
Generative Modelling of BRDF Textures from Flash Images
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Other Versions
A user study compares our approach favorably to previous work, even those with access to BRDF supervision. ...
Technically, we jointly embed all flash images into a latent space using a convolutional encoder, and -- conditioned on these latent codes -- convert random spatial fields into fields of BRDF parameters ...
propose a weakly supervised learning-based method for generating novel category-specific 3D shapes and demonstrate that it can help in learning material-class specific svBRDFs from images distributions ...
arXiv:2102.11861v2
fatcat:6me7cyt4jbhivbvpou3irpfuwi
« Previous
Showing results 1 — 15 out of 83 results