Bharath Raj

MS in Robotics @ Carnegie Mellon University | Advised by Dr. Shubham Tulsiani

me.jpg

I am interested in creating algorithms that can reason about scenes with the same proficiency as humans.

Everyday scenes consist of multiple objects positioned in diverse spatial arrangements, with each object presenting distinct physical properties, including dynamics, appearance and structure. Humans are able to implicitly create complex mental representations of these properties from just a brief glance, enabling them to plan and interact with the environment.

I am excited to research representations and techniques that can effectively leverage priors from large foundation models to infer physical properties of scenes from limited observations of casually captured images and videos towards enabling great applications in robotics and graphics.

Prior to joining CMU, I spent 3 incredible years at Siemens Digital Industries Software as part of the Intelligent Control Systems (ICS) team where I built models, algorithms and systems for myriad autonomous driving and general machine learning applications. Please checkout my resume for additional information about my work.

I am currently on the lookout for Summer Internship Opportunities starting May 2024! Please reach out if you think I would be a good fit for your team.

News

Dec 2023 Launched UpFusion, a method that can infer 3D representations of objects from as few as 1-6 images!
Aug 2022 Joined the MS in Robotics program at CMU!
Jul 2022 Left my role at Siemens Digital Industries Software after 3 wonderful years!

Selected Publications

  1. ITW_lock.gif
    UpFusion: Novel View Diffusion from Unposed Sparse View Observations
    Bharath Raj Nagoor Kani, Hsin-Ying Lee, Sergey Tulyakov, and Shubham Tulsiani
    arXiv preprint arXiv:2312.06661, 2023
  2. poster_human_pose.png
    Exploring Techniques to Improve Activity Recognition using Human Pose Skeletons
    Bharath Raj N., Anand Subramanian, Kashyap Ravichandran, and Venkateswaran N.
    2020 IEEE Winter Applications of Computer Vision Workshops (WACVW), 2020
  3. dehaze_img.png
    Single Image Haze Removal using a Generative Adversarial Network
    Bharath Raj N., and Venkateswaran N.
    2020 International Conference on Wireless Communications Signal Processing and Networking (WiSPNET), 2020

Selected Projects

  1. rendering_competition.png
    Progressive Photon Mapping
    Project Submission for Physics-based Rendering, 2023
  2. jetson.jpeg
    Deploying Tiny YOLOv2 on Jetson Nano using DeepStream
    Featured in Jetson Community Resources (Deep Learning section),