Experience
Research Fellow
[July 2019 - February 2020]
Working at MSR Bangalore with Dr Venkat Padmanabhan (Deputy Managing Director, MSR India) and Dr Sreangsu Acharyya (Senior Applied Scientist, MSR India).
I am using machine learning for a data-driven networking project at scale.
Undergraduate Thesis
[June 2018 - December 2018]
Completed my bachelor's thesis at the Augmented Vision Group, DFKI (German Research Centre for Artificial Intelligence) under Prof. Dr. Didier Stricker. The topic of my thesis was Deep Neural Networks for Scene Flow Estimation.
Worked in the Machine Vision and Autonomous Vehicles group with Dr. Oliver Wasenmüller and Mr René Schuster to model motion in cars. The project was funded by BMW.
Designed a novel architecture which achieved second-highest accuracy among end-to-end CNN methods for scene flow on the KITTI vision benchmark, with 48 times fewer parameters than the top-performing method.
This work was accepted for a full oral presentation at the 30th IEEE Intelligent Vehicles Symposium (IV'19) in Paris. Slides for the talk are available here. The paper is also available on arxiv. My thesis is available here.
Awarded the BITS Pilani Thesis Scholarship – 1 of 12 students among class of 1000 selected for the scholarship.
Co-Founder and Computer Vision Lead
[June 2018 - May 2019]
Sally Robotics is an 'Indian Autonomous Vehicles' research group by robotics researchers at the Centre for Robotics & Intelligent Systems, BITS Pilani and spearheaded by Prof Bijay Kumar Rout. The research group aims to develop a truly autonomous car for the challenging Indian Roads.
I'm currently associated with the development of the computer vision subsystem to achieve complete spatial perception for the Autonomous Vehicle.
Computer Vision Research Intern
[August 2017 - May 2018]
Using computer vision to detect traffic violations on dynamic Indian roads.
This project is based out of Multimedia Lab, IIT Delhi and is part of ITRA, Media Lab Asia. The project is being done under the guidance of Prof. Narendra Ahuja - Professor, UIUC and Director, ITRA.
Working on solving novel computer vision problems using deep learning, such as:
- Improving temporal association across frames for object detection and tracking
- Studying how to estimate motion (rotation and translation) of objects in video
- Exploring robust techniques to estimate ego-motion (motion of the body capturing the video)
Computer Vision Research Intern
[June - July 2017]
One of the 70 people selected globally to carry out a research project under the mentorship of Dr. Stephen Wolfram, founder of Wolfram Research and creator of Mathematica. The research group consisted of various PhD and graduate researchers.
My model will be incorporated in the Wolfram Neural Network Repository.
Researched deep neural network architectures for semantic segmentation of images in the Cityscapes Dataset at Wolfram Research, Waltham, USA. Click here for a report of my work.
The paper Pyramid Scene Parsing [CVPR 2017: Zhao et al.] was used as a baseline, which gives second highest accuracy on the Cityscapes Dataset. Various adjustments were made with the aim of improving accuracy and processing.
Selected for Google Summer of Code
[May - July 2017]
Multimodal emotion recognition using deep neural networks with research group at Red Hen Lab, University of Bremen, Germany as part of Google Summer of Code 2017.
Red Hen is a cooperative of engaged researchers who collaborate closely and contribute power and content to Red Hen and hence to each other and to future researchers.
Investigated the use of OpenPose for feature extraction from poses of individuals in images and video. Also investigated OpenSmile for feature extraction from audio. For merging the predictions from various modalities, the paper Video-based emotion recognition using CNN-RNN and C3D hybrid networks [Fan et al.], which was the winner of EmotiW 2016, was investigated.
Machine Learning Group
[August 2017 - May 2018]
Member of the machine learning group under Prof. Navneet Goyal, BITS Pilani.
- Investigating the role of convolutional layers in neural networks (which are primarily considered as "black boxes" in many applications)
- Achieved state-of-the-art accuracy with randomly initialised frozen convolution layers, suggesting the possibility that these work merely as input encoders in LeNet
- Extending this work onto other network architectures
Udacity Self-Driving Car Engineer Nanodegree
[December 2016 - March 2017]
One of the youngest students to be selected from amongst 11,000 applicants worldwide. The project will involve contributing to Udacity's real open source self-driving car (a 2016 Lincoln MKZ) after covering a detailed curriculum in deep learning, computer vision, sensor fusion, path planning and much more.
Currently completed a curriculum (lessons and projects) in deep learning and computer vision for self-driving cars.
Undergrad Robotics Researcher
[March 2016 - May 2018]
Youngest student to join research (in freshman year) in Embedded Systems & Robotics Lab, Computer Science Department, BITS Pilani.
Currently working on a project which aims to develop improvements for RTAB-Map. RTAB-Map is a RGBD Graph-Based SLAM approach based on an incremental appearance-based loop closure detector. This project investigates how recent advances in deep learning can be used to improve the performance in loop closure detection in SLAM.
Also working on implementing multi-robot SLAM.
Previously:
• Implemented various robot behaviours
• Written using Python, Embedded C, and Assembly
• Projects involved
- Lizi robot by RoboTiCan
- TurtleBot by ROS
- Firebird V by Nex Robotics
Core Team Member
[August 2015 - May 2019]
Senior member of the BITS Pilani Student Chapter of the Association for Computing Machinery (ACM).
Awarded as the best student chapter in India by ACM for the last 3 consecutive years. In 2017, ACM International also awarded us for Outstanding Recruitment from amongst student chapters all over the world.
Senior Member
[August 2015 - May 2019]
Involved with the Computer Science Association, BITS Pilani.
Conduct various events to get more people interested and help them out in issues related to Computer Science.
Member
[August 2015 - August 2016]
As part of the Department of Sponsorship and Marketing, worked to obtain sponsorship for Oasis, the cultural festival of BITS Pilani.
Oasis is the second largest college fest in India.
Debater
[August 2015 - August 2016]
Member of the Debating Society, BITS Pilani. Participated in and conducted Parliamentary style debates.