Tarasha Khurana

I work on perception problems for Optimus, Tesla's humanoid robot. I completed my PhD in Robotics at CMU with Prof. Deva Ramanan. My research focus is broadly on spatiotemporal 4D scene understanding and its applications.

Previously, I did my Masters in Robotics at CMU and my Bachelors in Computer Engineering at the University of Delhi. I've also spent some time at the Boston Dynamics AI Institute, Google Research, Argo AI, and Staqu Technologies.

I am a trained Kathak dancer and Hindustani classical vocalist. I love baking, taking care of plants, and hiking.

CV → Google Scholar → tkhurana@alumni.cmu.edu →

News

10/25

I started at Tesla. The energy around is electric!

09/25

I defended my PhD thesis! Recording is available here and draft is here.

09/25

Kaihua's work CogNVS was accepted at NeurIPS '25! He is looking for PhD positions.

06/25

Zihan's work on sparse-view reconstruction was accepted to ICCV '25!

06/25

I was selected for the CVPR 2025 Doctoral Consortium. A summary of my PhD work is here.

04/25

I gave a guest lecture in CMU's 16-720: Introduction to Computer Vision. Slides are here.

02/25

Kaihua's work on video amodal segmentation was accepted at CVPR '25!

12/24

Our workshop on 4D Vision: Modeling the Dynamic World was accepted at CVPR '25. I'm serving as the Program Chair.

03/24

We launched the 2nd iteration of the Argoverse2.0 Occupancy Forecasting Challenge.

02/24

I'll be at the Boston Dynamics AI Institute this summer.

08/23

I was chosen as a Rising Star in EECS!

07/23

I gave a talk on building 4D foundation models at Waabi AI.

06/23

We are all set to organize the Argoverse2.0 Challenges at CVPR 2023!

03/23

Our work on 4D occupancy forecasting was accepted to CVPR 2023!

07/22

Detecting Invisible People was covered by NBC Universal.

01/22

I'll be at Google Research this summer, working with Alireza Fathi and Cordelia Schmid!

07/21

My favorite work, Detecting Invisible People, was accepted to ICCV '21!

04/21

I accepted CMU's PhD offer for Fall '21.

Selected Work

Reconstruct, Inpaint, Finetune: Dynamic Novel-view Synthesis from Monocular Videos

Kaihua Chen*, Tarasha Khurana*, Deva Ramanan

NeurIPS ’25

pdf project page code

MonoFusion: Sparse-view 4D Reconstruction via Monocular Fusion

Zihan Wang, Jeff Tan, Tarasha Khurana*, Neehar Peri*, Deva Ramanan

ICCV ’25

pdf code

Using Diffusion Priors for Video Amodal Segmentation

Kaihua Chen, Deva Ramanan, Tarasha Khurana

CVPR ’25

pdf project page code

Predicting Long-horizon Futures by Conditioning on Geometry and Time

Tarasha Khurana, Deva Ramanan

arXiv 2404.11554, 2024

pdf project page

TAO-Amodal: A Benchmark for Tracking any Object Amodally

Cheng-Yen Hseih, Kaihua Chen, Achal Dave, Tarasha Khurana, Deva Ramanan

arXiv 2312.12433, 2023

pdf project page code

Point Cloud Forecasting as a Proxy for 4D Occupancy Forecasting

Tarasha Khurana, Peiyun Hu, David Held, Deva Ramanan

CVPR ’23

pdf project page code

Differentiable Raycasting for Self-supervised Planning

Tarasha Khurana*, Peiyun Hu*, Achal Dave, Jason Ziglar, David Held, Deva Ramanan

ECCV ’22

pdf project page code

BURST: A Benchmark for Unifying Object Recognition, Segmentation and Tracking in Videos

Ali Athar, Jonathon Luiten, Paul Voigtlaender, Tarasha Khurana, Achal Dave, Bastian Leibe, Deva Ramanan

WACV 2023

arXiv benchmark & API

Detecting Invisible People

Tarasha Khurana, Achal Dave, Deva Ramanan

ICCV ’21

arXiv project page code

TAO: A Large-Scale Benchmark for Tracking Any Object

Achal Dave, Tarasha Khurana, Pavel Tokmakov, Cordelia Schmid, Deva Ramanan

ECCV ’20 · Spotlight

arXiv project page workshop dataset