Pooja Ravi

Researcher at the University of Toronto

About me

I work with Professor David Lindell in the Toronto Computational Imaging Group (TCIG) on robot–object interaction and controllable video generation.

I completed my Master of Science in Applied Computing at the University of Toronto in 2025. I hope to pursue a PhD focused on shared action representations and meaningful sensory conditioning signals, with the goal of generating physically plausible action sequences that transfer across robotic embodiments.

Robot Manipulation
Action-conditioned World Models

Portrait of Pooja Ravi against a charcoal background

Publications

ProxyPose: 6-DoF Pose Tracking via Video-to-Video Translation
Ruihang Zhang*, Felix Taubner*, Pooja Ravi, Kiriakos N. Kutulakos, and David B. Lindell

NeurIPS · 2026

arXiv Code
Attention Mechanism, Linked Networks, and Pyramid Pooling Enabled 3D Biomedical Image Segmentation
Pooja Ravi, Srijarko Roy, Indira Dutta, and Kottilingam Kottursamy

IEEE/ACIS SNPD · 2022

Paper Code
Real-time Multi-Module Student Engagement Detection System
Pooja Ravi and Ali Dewan

ICCIS · Lecture Notes in Networks and Systems · 2022

Paper Code

Projects

Echo Charlie
Boson AI Hackathon winner · October 2025

A multimodal pipeline that restores speech to muted video using lip reading, transcript cleanup, and voice synthesis.

Project Page
2D Sketch to 3D Stylization
Turns grayscale sketches into stylized 3D views using neural radiance fields, few-shot learning, and CLIP-NeRF.

Code
MobileGen: Optimizing Diffusion Models
Optimizes diffusion models for mobile inference through quantization-aware training, distillation, and pruning.

Code