Kian Izadpanah — Computer Vision Researcher

01 — About

Who I Am

Hey, I'm Kian, a Computer Vision researcher who genuinely enjoys the process of figuring out how machines can understand and manipulate visual content. I finished my BSc in Computer Engineering at Sharif University of Technology in September 2025, and I've been lucky to work on some really interesting problems along the way.

Right now, I'm a postgraduate student in Computer Science at the University of Cyprus (UCY) while also collaborating with the Gruvi Lab at Simon Fraser University (SFU). My main focus these days is on Generative Models — it's a space that's moving incredibly fast, and I love being part of it.

I get to work with some truly great people. My supervisors — Ali Mahdavi-Amiri and Daniel Cohen-Or at SFU, and Yiorgos Chrysantou and Andreas Aristidou at UCY — have shaped the way I think about research more than I can say.

Location
Nicosia, Cyprus

Email
kianizadpanah@gmail.com

Focus
Generative Models

GitHub
KianIzadpanah

Things I care about

Computer Vision

Teaching machines to see — and actually understand what they're looking at

Generative Models

Diffusion models, Flux — anything that can create or transform visual content

Deep Learning

The underlying machinery that makes all of the above possible

02 — Education

Where I've Studied

Current

🇨🇾

University of Cyprus (UCY)

Computer Science

Feb. 2026 — Present

Doing my graduate research here, focused on generative models and computer vision. Great environment for the kind of work I want to do.

Supervisors: Y. Chrysantou & A. Aristidou

Collaboration

🇨🇦

Gruvi Lab @ Simon Fraser University

Research Collaboration

Sep. 2025 — Present

I collaborate remotely with this lab on image editing and generative model research. It's been a brilliant experience working with people at the top of the field.

Supervisors: A. Mahdavi-Amiri & D. Cohen-Or

Completed — Sep. 2025

Sharif University of Technology

BSc in Computer Engineering

Sep. 2021 — Sep. 2025

Sharif is where it all started for me. Four years of hard work, great professors, and some of the smartest classmates I've ever met. I took on graduate-level courses in ML and 3D Vision alongside the core curriculum — it pushed me in all the right ways.

3D Computer Vision ★ Machine Learning ★ Artificial Intelligence Linear Algebra Probability & Statistics Computer Simulation Game Theory Automata Theory Database Design

★ Graduate-level course

03 — Publications

What I've Published

A couple of papers I'm proud to have contributed to.

2026

Computer Vision · Robotics · Dataset

JRDB-Pose3D: A Multi-person 3D Human Pose and Shape Estimation Dataset for Robotics

Sandika Biswas, Kian Izadpanah, Hamid Rezatofighi

3D Human Pose SMPLX Robotics Dataset

2025

Networking · VANETs · Multimedia

CITRON: Collaborative Image Stitching and Redundancy Elimination for Energy-Efficient Multimedia Offloading in VANETs

Kian Izadpanah, Mohammad Hossein Soheilian, Bardia Safaei

Image Stitching VANETs Energy Efficiency Multimedia

04 — Research

Where I've Worked

Labs and groups I've had the privilege of being part of.

Jul. 2024 — Present

CORE Lab, Cambridge University

Research Assistant (Summer Internship)

Supervisor: Prof. Cengiz Öztireli

Working on generating and editing 3D human poses — it's a tricky problem with a lot of moving parts.
Getting hands-on with 3D frameworks like Joint2Human, FOF, SMPLX, STAR, and SPIN.
Using text-driven models like CLIP, CLIP-PAE, and CLIP-Forge to guide generation.

Nov. 2023 — Present

VL4AI Lab, Monash University

Research Assistant

Supervisor: Prof. Hamid Rezatofighi

Contributing to the JRDB dataset — specifically improving 3D human pose estimation and correction at scale.
Working with SMPLX for 3D mesh generation and Quaternion representations for rotations.

Jul. 2023 — Nov. 2023

Robust/Interpretable ML Lab, Sharif University

Research Assistant

Supervisor: Prof. Mohammad Hossein Rohban

Worked on making a medical detection model generalise better across hospitals and scanners — the classic domain adaptation challenge, but with real stakes.
Combined diffusion-based generative models and a VAE-based OOD detector to boost RetinaNet's performance on the MIDOG benchmark.

Mar. 2023 — Jun. 2023

Google

Research Assistant

Supervisor: Prof. Hossein Mobahi

Built AGSAM (Adaptive-Gradient SAM) — an improved version of Sharpness-Aware Minimization that adapts to the geometry of the loss landscape.
Extended the core gradient idea by factoring in both the slope and the curvature of the objective function.

Feb. 2023 — Apr. 2023

Robust/Interpretable ML Lab, Sharif University

Research Assistant

Supervisor: Prof. Mohammad Hossein Rohban

Dug into anomaly detection and backdoor attacks — how they work, and how to defend against them.
Applied OOD scoring techniques to video surveillance settings, which was a fun challenge in itself.

05 — Skills