Tavi Halperin

Computer Vision and Deep Learning Researcher @ Lightricks
tavihalperin@gmail
Google Scholar   |    Linkedin   |    Github   

As a Researcher and Team Lead at the research department of Lightricks, my primary focus lies on state-of-the-art generative image and video models, as well as their practical applications.
Previously, I completed a PhD in Computer Science at the Hebrew University of Jerusalem. My thesis focused on video processing, geometry, and the use of deep learning to simultaneously learn from audio and video. My advisor was Prof. Shmuel Peleg

Publications

CAFA: a Controllable Automatic Foley Artist

Roi Benita*, Michael Finkelson*, Tavi Halperin, Gleb Sterkin, Yossi Adi

arXiv 2025

arXiv   project   code  

V-LASIK: Consistent Glasses-Removal from Videos Using Synthetic Data

Rotem Shalev-Arkushin, Aharon Azulay, Tavi Halperin, Eitan Richardson, Amit H. Bermano, Ohad Fried

International Conference on Learning Representations (ICLR) Workshop 2025

arXiv   project   code  

Diffusing Colors: Image Colorization with Text Guided Diffusion

Nir Zabari, Aharon Azulay, Alexey Gorkor, Tavi Halperin, Ohad Fried

ACM Transactions on Graphics (Proc. SIGGRAPH Asia), 2023

arXiv   project  

Temporally stable video segmentation without video annotations

Aharon Azulay, Tavi Halperin , Orestis Vantzos, Nadav Bornstein, Ofir Bibi

IEEE Winter Conference on Applications of Computer Vision (WACV) 2022

arXiv   video  

Endless Loops: Detecting and Animating Periodic Patterns in Still Images

Tavi Halperin, Hanit Hakim, Orestis Vantzos, Gershon Hochman, Gal Nachmana, Netai Benaim, Lior Sassy, Michael Kupchik, Ofir Bibi, Ohad Fried

ACM Transactions on Graphics (Proc. SIGGRAPH), 2021

project   arXiv   video   supplementary   mobile app  blogpost 

Neural separation of observed and unobserved distributions

Tavi Halperin, Ariel Ephrat and Yedid Hoshen

International Conference on Machine Learning (ICML) 2019

arXiv   code  

Dynamic Temporal Alignment of Speech to Lips

Tavi Halperin*, Ariel Ephrat* and Shmuel Peleg

IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) 2019

arXiv   code   video  

[*equal contribution]

Clear Skies Ahead: Towards Real-Time Automatic Sky Replacement in Video

Tavi Halperin, Harel Cain, Ofir Bibi and Michael Werman

Eurographics 2019

arXiv   video  

Seeing through noise: Speaker separation and enhancement using visually-derived speech

Aviv Gabbay, Ariel Ephrat, Tavi Halperin and Shmuel Peleg

IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) 2018

arXiv   project   video  

An Epipolar Line from a Single Pixel

Tavi Halperin and Michael Werman

IEEE Winter Conference on Applications of Computer Vision (WACV) 2018

arXiv   video  

Improved Speech Reconstruction from Silent Video

Ariel Ephrat*, Tavi Halperin* and Shmuel Peleg

ICCV Workshop on Computer Vision for Audio-Visual Media 2017

arXiv   project   video  

[*equal contribution]

Egosampling: Wide view hyperlapse from egocentric videos

Tavi Halperin, Yair Poleg, Chetan Arora and Shmuel Peleg

IEEE Transactions on Circuits and Systems for Video Technology (TCSVT) 2017

pdf  

Epipolar geometry based on line similarity

Gil Ben-Artzi, Tavi Halperin, Michael Werman and Shmuel Peleg

International Conference on Pattern Recognition (ICPR) 2016

pdf  

EgoSampling: Fast-Forward and Stereo for Egocentric Videos

Yair Poleg, Tavi Halperin, Chetan Arora and Shmuel Peleg

IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2015

pdf   project   video  

Patents

Method and system for enhancing a speech signal of a human speaker in a video using visual information

Shmuel Peleg, Asaph Shamir, Tavi Halperin, Aviv Gabbay, Ariel Ephrat

U.S. Patent 10,475,465

Method and system for generating adaptive fast forward of egocentric videos‏

Shmuel Peleg, Yair Poleg, Tavi Halperin and Chetan Arora

U.S. Patent 9,672,626