SharinGAN: Combining Synthetic and Real Data for Unsupervised Geometry Estimation

Pnvr, Koutilya; Zhou, Hao; Jacobs, David W.

doi:10.1109/cvpr42600.2020.01399

Cited by 14 publications

(2 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The authors of paper [3] introduced another approach for unsupervised learning. They used a Generative Adversarial Network (GAN) for the generation of synthetic data and combined these data with real images during the training process to determine the geometric information from a single image.…”

Section: Related Workmentioning

confidence: 99%

Lightweight 2D Map Construction of Vehicle Environments Using a Semi-Supervised Depth Estimation Approach

Kashevnik

Ali

2023

Intels’22

View full text Add to dashboard Cite

show abstract

Section: Related Workmentioning

confidence: 99%

Lightweight 2D Map Construction of Vehicle Environments Using a Semi-Supervised Depth Estimation Approach

Kashevnik

Ali

2023

Intels’22

View full text Add to dashboard Cite

show abstract

“…Another approach, gaining more attention due to a recent renaissance of generative models, is to generate large amounts of realistic synthetic images for use as training data [40]. Recent trends are to mix synthetic and actual data [22], e.g. pretrain a model on a larger synthetically-generated corpus, and fine-tune on a smaller actual sample set [41], or guide the generation of synthetic data to make it more domainspecific [17].…”

Section: Related Workmentioning

confidence: 99%

Human-Aided Saliency Maps Improve Generalization of Deep Learning

Boyd¹,

Bowyer²,

Czajka³

2021

Preprint

View full text Add to dashboard Cite

Deep learning has driven remarkable accuracy increases in many computer vision problems. One ongoing challenge is how to achieve the greatest accuracy in cases where training data is limited. A second ongoing challenge is that trained models are sometimes fragile in the sense that the accuracy achieved does not generalize well, even to new data that is subjectively similar to the training set. We address these challenges in a novel way, with the first-ever (to our knowledge) exploration of encoding human judgement about salient regions of images into the training data. We compare the accuracy and generalization of a state-of-the-art deep learning algorithm for a difficult problem in biometric presentation attack detection when trained on (a) original images with typical data augmentations, and (b) the same original images transformed to encode human judgement about salient image regions. The latter approach results in models that achieve higher accuracy and better generalization, decreasing the error of the LivDet-Iris 2020 winner from 29.78% to 16.37%, and achieving impressive generalization in a leave-one-attacktype-out evaluation scenario. This work opens a new area of study for how to embed human intelligence into training strategies for deep learning to achieve high accuracy and generalization in cases of limited training data.

show abstract

PatchRefiner: Leveraging Synthetic Data for Real-Domain High-Resolution Monocular Metric Depth Estimation

Li,

Bhat,

Wonka

2024

Lecture Notes in Computer Science

View full text Add to dashboard Cite

SharinGAN: Combining Synthetic and Real Data for Unsupervised Geometry Estimation

Cited by 14 publications

References 30 publications

Lightweight 2D Map Construction of Vehicle Environments Using a Semi-Supervised Depth Estimation Approach

Lightweight 2D Map Construction of Vehicle Environments Using a Semi-Supervised Depth Estimation Approach

Human-Aided Saliency Maps Improve Generalization of Deep Learning

PatchRefiner: Leveraging Synthetic Data for Real-Domain High-Resolution Monocular Metric Depth Estimation

Contact Info

Product

Resources

About