Ayan Sinha
YOU?
Author Swipe
View article: Depth Estimation by Learning Triangulation and Densification of Sparse Points for Multi-view Stereo
Depth Estimation by Learning Triangulation and Densification of Sparse Points for Multi-view Stereo Open
Multi-view stereo (MVS) is the golden mean between the accuracy of active depth sensing and the practicality of monocular depth estimation. Cost volume based approaches employing 3D convolutional neural networks (CNNs) have considerably im…
View article: DELTAS: Depth Estimation by Learning Triangulation And densification of\n Sparse points
DELTAS: Depth Estimation by Learning Triangulation And densification of\n Sparse points Open
Multi-view stereo (MVS) is the golden mean between the accuracy of active\ndepth sensing and the practicality of monocular depth estimation. Cost volume\nbased approaches employing 3D convolutional neural networks (CNNs) have\nconsiderably…
View article: Efficient 2.5D Hand Pose Estimation via Auxiliary Multi-Task Training\n for Embedded Devices
Efficient 2.5D Hand Pose Estimation via Auxiliary Multi-Task Training\n for Embedded Devices Open
2D Key-point estimation is an important precursor to 3D pose estimation\nproblems for human body and hands. In this work, we discuss the data,\narchitecture, and training procedure necessary to deploy extremely efficient\n2.5D hand pose es…
View article: Efficient 2.5D Hand Pose Estimation via Auxiliary Multi-Task Training for Embedded Devices
Efficient 2.5D Hand Pose Estimation via Auxiliary Multi-Task Training for Embedded Devices Open
2D Key-point estimation is an important precursor to 3D pose estimation problems for human body and hands. In this work, we discuss the data, architecture, and training procedure necessary to deploy extremely efficient 2.5D hand pose estim…
View article: Imaging through glass diffusers using densely connected convolutional networks
Imaging through glass diffusers using densely connected convolutional networks Open
Computational imaging through scatter generally is accomplished by first characterizing the scattering medium so that its forward operator is obtained and then imposing additional priors in the form of regularizers on the reconstruction fu…
View article: Gradient Adversarial Training of Neural Networks
Gradient Adversarial Training of Neural Networks Open
We propose gradient adversarial training, an auxiliary deep learning framework applicable to different machine learning problems. In gradient adversarial training, we leverage a prior belief that in many contexts, simultaneous gradient upd…
View article: Imaging through glass diffusers using densely connected convolutional networks
Imaging through glass diffusers using densely connected convolutional networks Open
Computational imaging through scatter generally is accomplished by first characterizing the scattering medium so that its forward operator is obtained; and then imposing additional priors in the form of regularizers on the reconstruction f…
View article: Lensless computational imaging through deep learning
Lensless computational imaging through deep learning Open
Deep learning has been proven to yield reliably generalizable solutions to numerous classification and decision tasks. Here, we demonstrate for the first time to our knowledge that deep neural networks (DNNs) can be trained to solve end-to…
View article: Severe asthma in humans and mouse model suggests a CXCL10 signature underlies corticosteroid-resistant Th1 bias
Severe asthma in humans and mouse model suggests a CXCL10 signature underlies corticosteroid-resistant Th1 bias Open
We previously showed that Th1/type 1 inflammation marked by increased IFN-γ levels in the airways can be appreciated in 50% of patients with severe asthma, despite high dose corticosteroid (CS) treatment. We hypothesized that a downstream…
View article: SurfNet: Generating 3D shape surfaces using deep residual networks
SurfNet: Generating 3D shape surfaces using deep residual networks Open
3D shape models are naturally parameterized using vertices and faces, \ie, composed of polygons forming a surface. However, current 3D learning paradigms for predictive and generative tasks using convolutional neural networks focus on a vo…
View article: Deconvolving Feedback Loops in Recommender Systems
Deconvolving Feedback Loops in Recommender Systems Open
Collaborative filtering is a popular technique to infer users' preferences on new content based on the collective information of all users preferences. Recommender systems then use this information to make personalized suggestions to users…
View article: Lensless computational imaging through deep learning
Lensless computational imaging through deep learning Open
Deep learning has been proven to yield reliably generalizable answers to numerous classification and decision tasks. Here, we demonstrate for the first time, to our knowledge, that deep neural networks (DNNs) can be trained to solve invers…
View article: Physics Based Supervised and Unsupervised Learning of Graph Structure
Physics Based Supervised and Unsupervised Learning of Graph Structure Open
Graphs are central tools to aid our understanding of biological, physical, and social systems. Graphs also play a key role in representing and understanding the visual world around us, 3D-shapes and 2D-images alike. In this dissertation, I…