Profile Pic
Utsav Garg Computer Science Undergrad, Budding Researcher & Developer

I am a Computer Science B.Eng. student with a focus in Intelligent Systems at Nanyang Technological University, Singapore. Currently, I am working on applying deep learning approaches to the task of action recognition for my final year project under the supervision of Jagath C Rajapakse. My research interest lies in solving computer vision problems by using deep learning approaches and I am always on the lookout for opportunities to learn more and work on exciting projects in this field. I also actively maintain Fabrik, an open source project aimed at lowring the barrier for getting started with deep learning. In my spare time I enjoy listening to music and playing the electronic keyboard.

Research

Fast Feature Fool: A data independent approach to universal adversarial perturbations

Konda Reddy Mopuri*, Utsav Garg*, R. Venkatesh Babu

British Machine Vision Conference(BMVC), 2017

The paper presents a novel data independent approach to generate image agnostic adversarial perturbations for a range of CNNs trained for object recognition. It further show that these perturbations are transferable across multiple network architectures trained either on the same or different data.





CNN Fixations: An unraveling approach to visualize the discriminative image regions

Konda Reddy Mopuri*, Utsav Garg*, R. Venkatesh Babu

arXiv

The paper presents an approach to provide visual explanations for the network’s predictions. Our approach can analyse a variety of CNN based models trained for vision applications such as object recognition and caption generation.





NAG: Network for Adversary Generation

Konda Reddy Mopuri*, Utkarsh Ojha*, Utsav Garg, R. Venkatesh Babu

Computer Vision and Pattern Recognition(CVPR), 2018

The paper presents a novel generative approach to model the distribution of adversarial perturbations. Our trained generator network attempts to capture the distribution of adversarial perturbations for a given classifier and readily generates a wide variety of such perturbations.




(* = equal contribution)

Experience
  • November 2017 - January 2018
    Google Code-in

    Mentor, CloudCV

    Singapore

    Mentored over 20 students for CloudCV, helping them with various development problems, leading to the successful completion of over 50 tasks.

  • May 2017 - August 2017
    Google Summer of Code

    Student Developer, CloudCV

    India/Singapore

    Worked on and currently maintain Fabrik, an open-source web application to collaboratively build, visualise and design neural networks.

  • August 2016 - May 2017
    Video Analytics Lab

    Research Intern

    IISc, Bangalore, India

    Worked on multiple research problems which apply deep learning approaches to computer vision tasks, particularly related to understanding CNNs and analysing the effects of adversarial images.

  • May 2015 - July 2015
    Newgen Software

    Project Intern

    New Delhi, India

    Developed a web application to create custom forms in which the appearance and location of each field can be customized. Intended to generate forms that replace the conventional forms used in financial institutions

Selected Projects

Fabrik

An open source web application to collaboratively build and visualize neural networks with a simple drag-and-drop interface.


Slides In Time

A web application that can build slides for you as you speak. No need to spend hours on formatting and preparing your presentation.


TF Classification

A unified program to perform image classification for 8 popular convolutional neural networks through a single interface



Drone Gesture Control

A HCI that uses hand gestures to simultaneously control multiple quad-copters along with a touch based UI for manual controls and video feed.


Maze Solving Robot

Algorithm for a maze solving robot, the robot first needs to explore the maze and then take the fastest path from the start to goal cell.


Sketch SVG

An android application with a drawing canvas with an option to save the drawings in SVG format.


Education
  • August 2014 - Present
    Nanyang Technological University

    Bachelor of Engineering in Computer Science with Focus in Intelligent Systems

    Singapore

  • January 2016 - May 2016
    University of Waterloo

    Exchange Program (Computer Science)

    Waterloo, Canada

  • April 2010 - May 2014
    La Martiniere College

    High School and Higher Secondary School

    Lucknow, India