Rajat Sahay

I am a final year Master's student at RIT, majoring in Data Science.

I am currently working as a Graduate Research Assistant at the RIT Center for Human Aware AI under the supervision of Prof. Andreas Savakis. My work involves understanding representation learning in visual models through the lens of unsupervised domain adaptation.

I have had the pleasure of conducting research alongside researchers and scientists across the world. Prior to this, I have been associated with the Juno Mission at NASA JPL, IRIS-HEP at Princeton, the LIVIA Lab at ETS Montreal, ULR France, CamCann Smart Systems, and IIT Indore.

Email  /  CV  /  Google Scholar  /  Blog  /  Twitter  /  Github  /  LinkedIn

profile photo
Research

I'm interested in computer vision, machine learning, perceptual interfaces, and image processing. Much of my research is about inferring the physical world (shape, motion, color, light, etc) from images and videos. I also like to focus on interdisciplinary research, applying vision-based methods to innovatiely tackle problems in different fields. More recently, I have been focusing on problems at the intersection of computer vision and AI fairness, ethics and algorithmic biases as well as their applications in the real world.

Image Unavailable Dynamic Template Selection through Change Detection for Adaptive Siamese Tracking
Madhu Kiran, Le Thanh Nguyen-Meidine, Rajat Sahay, Rafael Menelau-Cruz, Louis-Antoine Blais-Morin, Eric Granger

IJCNN 2022 (Oral Presentation)
arXiv/paper

We formulate Single Object Tracking (SOT) as an online incremental learning problem and develop a new method for dynamic sample selection and memory replay, preventing template corruption.

Image Unavailable Generative Target Update for Adaptive Siamese Tracking
Madhu Kiran, Le Thanh Nguyen-Meidine, Rajat Sahay, Rafael Menelau-Cruz, Louis-Antoine Blais-Morin, Eric Granger

ICPRAI 2022 (Oral Presentation)
arXiv/paper/project/code

We propose a model adaptation method for Siamese trackers. We employ a generative model to produce a synthetic template from the object search regions of several previous frames, rather than directly using the tracker output.

Image Unavailable Graph Segmentation in Scientific Datasets
Rajat Sahay, Savannah Thais

NeurIPS (Workshop) 2021
paper/poster

We explore graph segmentation as a precursor to a larger deep learning pipelines, improving accuracy and efficiency of downstream tasks by using the underlying geometric structure of data. We apply our research to track particles generated during proton-proton collisions in the Large Hadron Collider (LHC).

clean-usnob An Enhanced Prototypical Network Architecture for Few-Shot Handwritten Urdu Character Recognition
Rajat Sahay, Mickael Coustaty

IEEE Access
paper/arXiv

We enhance Prototypical Networks with additional self-supervised modules to learn Euclidian embeddings and classify handwritten Urdu characters using a minimal number of examples.

clean-usnob Unrestricted Adversarial Attacks on Vision Transformers
Rajat Sahay

CVPR (Workshop) 2021
paper/arXiv/poster

We leverage colorization models to semantically attack images generating low-frequency adversarial perturbations while changing colors in ambigous areas. We were able to demonstrate that this method was successful in getting Vision Transformers (ViTs) to misclassify image labels.


Image Unavailable Visual Odometry Problems in Constrained Environments
Rajat Sahay, Surya Prakash
[Project Report]

Leveraged optical flow methods to map trajectories of random, fast-moving particles in constrained environments. Used statistical methods like Lucas-Kanade Optical Flow combined with Hungarian Algorithm to track and predict feature frames based on inputs by previous frames.

Updates

Jun '22

Dynamic Template Selection through Change Detection for Adaptive Siamese Tracking got accepted to IEEE WCCI and IJCNN'22.

Mar '22

Generative Target Update for Adaptive Siamese Tracking got accepted to ICPRAI'22.

Oct '21

Graph Segmentation in Scientific Datasets got accepted to the ML4PS Workshop at NeurIPS'21.

Oct '21

I'll be continuing my work with Dr. Glenn Orton at the Planetary and Exoplanetary Systems Department at NASA Jet Propulsion Laboratory from January 2022.

Jun '21

Unrestricted Adversarial Attacks on Vision Transformers got accepted to the AML-CV Workshop at CVPR'21.

May '21

Super excited to be selected for the JPL Visiting Student Research Program. I'll be working with Dr. Glenn Orton at the Planetary and Exoplanetary Systems Department at NASA Jet Propulsion Laboratory in September.

Mar '21

I'll be working with Dr. Savannah Thais at the IRIS-HEP Software Institute, Princeton University on applying ML methods to solve High Energy Physics problems.

Dec '20

I'll be working with Prof. Mickael Coustaty and Prof. Jean-Loup Guillaume at the L3i Laboratoire, University of La Rochelle, France.

Dec '20

Excited to be a recipient of the Mitacs Globalink Resarch Scholarship! I would be interning with Prof. Eric Granger at the LIVIA Laboratory, ETS Montreal, Canada over the summer.

Jun '20

Due to the pandemic, I would be remotely beginning my role as a Research Intern with Prof. Mickael Coustaty at the L3i Laboratoire, University of La Rochelle, France.

Jan '20

I'll be beginning my work at CamCann Smart Systems as a Computer Vision Engineer Intern. Looking forward to working with the insanely talented team.

May '19

I'll be starting my internship with Prof. Surya Prakash at the Pattern Analysis and Machine Intelligence Laboratory, Indian Institute of Technology, Indore, India.

[ Fun Website ] [ Template ]