Alexander 'Sasha' Vilesov

I am a PhD student in the Visual Machines Group with Prof. Achuta Kadambi at UCLA where I started in 2021. At VMG, we work on developing cutting-edge tools at the intersection of physics and artificial intelligence, to be applied to diverse problems in human-centric computer vision and computational imaging.

Before starting my PhD, I received my Bachelor's in Electrical and Computer Engineering at the University of Southern California (USC) in 2021 and worked in the GSP Lab with Prof. Antonio Ortega and the S2L2 Lab on inverse reinforcement learning with Prof. Rahul Jain . In 2020-2021, I worked at NASA JPL on TriG GPS receivers (mounted on Cosmic-2) to support tracking of GALILEO satellites.

Email  /  CV  /  Twitter  /  Github

vilesov@ucla.edu  /  (626)-390-7241

profile photo
Research

My research interests lie at the intersection computer vision/computational imaging and human-centric machine learning. My current work spans 3D representations, equitable health sensing, and uncertainty.

Papers

* indicates equal contribution

Implicit Neural Models to Extract Heart Rate from Video
Pradyumna Chari, Anirudh B. Harish, Adnan Armouti, Alexander Vilesov, Sanjit Sarda Laleh Jalilian, Achuta Kadambi
ECCV, 2024  
Project Page / Paper Link / Code

We propose a new implicit neural representation, that enables fast and accurate decomposition of face videos into blood and appearance components. This allows contactless estimation of heart rate from challenging out-of-distribution face videos.

Solutions to Deepfakes: Can Camera Hardware, Cryptography, and Deep Learning Verify Real Images?
Alexander Vilesov, Yuan Tian, Nader Sehatbakhsh, Achuta Kadambi
White Paper, 2024  
Paper Link

The rapid advancement of generative AI threatens the credibility of real images and videos, as synthetic content will soon be indistinguishable from camera-captured media and easily accessible to all. This white paper explores detection and cryptographic methods to reliably differentiate real images from synthetic ones, analyzing existing strategies and proposing improvements to enhance their effectiveness.

CG3D: Compositional Generation for Text-to-3D via Gaussian Splatting
Alexander Vilesov*, Pradyumna Chari*, Achuta Kadambi
Arxiv, 2023  
Project Page / Paper Link

We present a method for 3D generation of multi-object realistic scenes from text by utilizing text-to-image diffusion models and Gaussian radiance fields. These scenes are decomposable and editable at the object level.

Making thermal imaging more equitable and accurate: resolving solar loading biases
Ellin Zhao, Alexander Vilesov, Shreeram Athreya, Pradyumna Chari, Jeanette Merlos, Kendall Millett, Nia St Cyr, Laleh Jalilian, Achuta Kadambi
Arxiv, 2023  
Paper Link

Despite the wide use of thermal sensors for temperatures screening, estimates from thermal sensors do not work well in uncontrolled scene conditions such as after sun exposure. We propose a single-shot correction scheme to eliminate solar loading bias in the time of a typical frame exposure (33ms).

Blending Camera and 77 GHz Radar Sensing for Equitable, Robust Plethysmography
Alexander Vilesov*, Pradyumna Chari*, Adnan Armouti*, Anirudh B. Harish, Kimaya Kulkarni, Ananya Deoghare, Laleh Jalilian, Achuta Kadambi
SIGGRAPH, 2022  
Project Page / Paper Link / Code

To overcome fundamental skin-tone biases in camera-based remote plethysmography, we propose an adversarial learning-based fair fusion method, using a novel RGB Camera and FMCW Radar hardware setup.