Aditya Vora

I am a Ph.D student at the GrUVi Lab of School of Computer Science at Simon Fraser University, under the supervision of Prof. Hao (Richard) Zhang.

Prior to that, I spent couple of years working in industry where I was mainly involved in development of computer vision and machine learning solutions for fire safety and security products. I completed my master's from Indian Institute of Technology, Gandhinagar, India where I did my thesis under the supervision of Prof. Shanmuganathan Raman. I received my bachelor's degree in Electronics and Communications Engineering from Birla Vishwakarma Mahavidyalaya, India.

Email  /  Google Scholar  /  LinkedIn  /  GitHub  /  Twitter

profile photo
Research

My research interests are in 3D Computer Vision and Graphics, Geometry Processing and Deep Learning.

Publications

DiViNeT: 3D Reconstruction from Disparate Views using Neural Template Regularization
Aditya Vora, Akshay Gadi Patil, Hao (Richard) Zhang
Neural Information Processing Systems (NeurIPS), 2023
ArXiv / Project Page / code / bibtex

We present a volume rendering-based neural surface reconstruction method that takes as few as three disparate RGB images as input. Our key idea is to regularize the reconstruction, which is severely ill-posed and leaving significant gaps between the sparse views, by learning a set of neural templates to act as surface priors.

Deep Appearance Consistent Human Pose Transfer
Ashish Tiwari, Zeeshan Khan, Aditya Vora, Manjuprakash Rama Rao, Shanmuganathan Raman
International Conference of Pattern Recognition (ICPR), 2022
Paper / bibtex

We present a robut deep architecture for Appearance Consistent person image generation in novel poses. We incorporate a 3 stream network, for image, pose, and appearance. Additionaly we use Gated convolutions and, Non-local attention blocks for generating realistic images.

Iterative spectral clustering for unsupervised object localization
Aditya Vora, Shanmuganathan Raman
Pattern Recognition Letters (PRL), 2018
arxiv / bibtex

We propose a completely unsupervised algorithm for the same, where we try to exploit the structural differences between the foreground and the background region in an image in order to localize the object in the scene.

Flow-Free Video Object Segmentation
Aditya Vora, Shanmuganathan Raman
National Conference on Computer Vision, Pattern Recognition, Image Processing and Graphics (NCVPRIPG), 2017
arxiv / bibtex

We propose an fully automatic video object segmentation algorithm where localization of object segments are obtained by performing clustering on proposals generated by a segmentation proposal network. Later, the overall temporal consistency is improved using a track and fill method.

(Pr)ePrints

FCHD: Fast and accurate head detection in crowded scenes
Aditya Vora, Vinay Chilaka
arxiv / bibtex / code

A fully convolutional single stage head detector is proposed where the anchor scales are designed by taking effective receptive field into account, hence giving better average precision especially for small heads in crowded scenes.

Patents

Rajkumar Palanivel, Amit Kulkarni, Douglas Beaudet, Manjuprakash Rama Rao, Atul Laxman Katole, Aditya Vora, "System and Method for identifying blockages of emergency exits in a building.", US Patent 11388775, 2021

Teaching

TA - CMPT 985: Neural Fields by Prof. Andrea Tagliasacchi SFU [Summer, 2023]

TA - CMPT 361: Introduction to Visual Computing by Prof. Jason Peng SFU [Spring, 2023]

Service

Reviewer

CVPR 2024



Layout inspired by Jon Barron . Thank you Jon. © 2020