Experience

• 

  Indian institute of technology, delhi

  May 2019 - May 2023

  Undergraduate student
• 

  Board for student welfare (bsw), iit delhi

  Dec 2020 - Feb 2021

  Academic mentor
• 

  Torch investment group

  Jun 2021 - Oct 2021

  Quantitative strategist
• 

  Mastercard

  Jun 2022 - Jul 2022

  Machine learning intern
• 

  University of maryland

  Aug 2023 - now

  Masters student
• 

  Teli’s research lab

  Dec 2023 - May 2024

  Research assistant

  Accurate detection and understanding of object motion and direction in videos is crucial for various computer vision tasks like autonomous driving, navigation, etc. Worked on a novel deep learning approach that leverages FlowFormer++ and YOLOv8 models to extract detailed motion information for objects in video sequences. FlowFormer++ estimates optical flow between consecutive frames, capturing pixel-wise motion information. YOLOv8 identifies and segments objects within each frame. By combining these outputs, the system calculates the average flow within each object mask, providing motion direction and 2D velocity. Additionally, monocular depth estimation is utilized to estimate the depth and, subsequently, the 3D velocity of each object in the scene. This comprehensive approach offers a richer understanding of object motion in videos, enabling applications in autonomous systems and real-time scene analysis. Show less

• 

  Mirage lab

  Jun 2024 - now

  Research assistant