GeezerVison

Designing adaptive spaces with computer vision, generative AI, and mixed reality.

I am an Assistant Professor of Civil and Environmental Engineering and, by courtesy, of Computer Science at Stanford University. I lead the Gradient Spaces group, researching at the intersection of Computer Vision, Generative AI, and the built environment. My work focuses on developing machine perception architectures and applications capable of designing, reconstructing, and predicting in adaptive physical and digital spaces.

Building intelligent systems for the spaces people inhabit.

My area of focus is on developing quantitative and data-driven methods that learn from real-world visual data to generate, predict, and simulate new or renewed built environments that place the human in the center.

My goal is to create sustainable, inclusive, and adaptive built environments that can support our current and future physical and digital needs.

I am intrigued by the idea of creating responsive spaces that blend from the 100% physical (real reality) to the 100% digital (virtual reality) and anything in between, with the use of Mixed Reality.

Current teaching

Designing for Gradient Spaces — CEE342, Stanford, Spring [Website]

Computer Vision for the Built Environment — CEE 247C, Stanford, Winter [Website]

AI Applications in AEC — CEE 329, Stanford, Spring [Website]

Lab themes

Spatial intelligence

Perception systems for understanding real-world spaces in 3D and 4D, with emphasis on semantic scene understanding and geometric reasoning.

Generative design

Learning models that synthesize and predict spaces, support planning, and reason over future built conditions.

Adaptive environments

Designing responsive, data-driven physical and digital spaces that support sustainability, resilience, and human-centered use.

Academic publications

Search publications

Recent publications

Automated Facade Solar Retrofits Using Feed-forward 3D Reconstruction and Generative Image Editing

Ayca Duran, Christoph Waibel, Bernd Bickel, Iro Armeni, Arno Schlueter

Automation in Construction, Vol. 193, 107294, 2027

PDF

FAMOS: Feed-Forward 3D Articulation Modeling from Sparse Observations

Kevin Qu, Tao Sun, Massimiliano Viola, Liyuan Zhu, Zhizhuo Zhou, Sayan Deb Sarkar, Konrad Schindler, Iro Armeni

WACV 2027

PDF Website

SceneKey: Multi-modal Embedding for Fine-grained 3D Scene Retrieval

Luka Milivojevic, X. Wang, Y. Miao, Sayan Deb Sarkar, L. Sun, Nikola Popovic, Iro Armeni, Danda Pani Paudel

WACV 2027

World from Motion: Generative Dynamic Gaussian Reconstruction from Monocular Video

Liyuan Zhu, Shengyu Huang, Amrita Mazumdar, Tianye Li, Zan Gojcic, Gordon Wetzstein, Iro Armeni, Shalini De Mello, Alex Trevithick

NeurIPS 2026

PDF Website

GraphWrit3R: End-to-End Writing of Scene Graphs for Multi-Modal 3D Scenes

Luka Milivojevic, Nikola Popovic, Sayan Deb Sarkar, Sebastian Koch, Iro Armeni, Luc Van Gool, Danda Pani Paudel

NeurIPS 2026

CoPE-VideoLM: Leveraging Codec Primitives For Efficient Video Language Modeling

Sayan Deb Sarkar, Rémi Pautrat, Ondrej Miksik, Marc Pollefeys, Iro Armeni, Mahdi Rad, Mihai Dusmanu

NeurIPS 2026

GeezerVison

Designing adaptive spaces with computer vision, generative AI, and mixed reality.

Building intelligent systems for the spaces people inhabit.

Background and recognition

Education

Awards

Current teaching

Lab themes

Spatial intelligence

Generative design

Adaptive environments

Academic publications

Recent publications

Automated Facade Solar Retrofits Using Feed-forward 3D Reconstruction and Generative Image Editing

FAMOS: Feed-Forward 3D Articulation Modeling from Sparse Observations

SceneKey: Multi-modal Embedding for Fine-grained 3D Scene Retrieval

World from Motion: Generative Dynamic Gaussian Reconstruction from Monocular Video

GraphWrit3R: End-to-End Writing of Scene Graphs for Multi-Modal 3D Scenes

CoPE-VideoLM: Leveraging Codec Primitives For Efficient Video Language Modeling

Effective Multi-sensor Conditioning for Street-view Novel-view Synthesis

Energy-based Compositional Diffusion Planning

Leveraging vision and language models to capture residential seismic vulnerabilities in the building inventory

Register Any Point: Scaling 3D Point Cloud Registration by Flow Matching

ReScene4D: Temporally Consistent Semantic Instance Segmentation of Evolving Indoor 3D Scenes

GaussFusion: Improving 3D Reconstruction in the Wild with Geometry-Informed Video Generators

WildPose: A Unified Framework for Robust Pose Estimation in the Wild

Urban-scale Facade Material Mapping from Street View Images Using Vision–Language Models for Circular Construction Planning

Deep Sketch-Based 3D Modeling: A Survey

Do 3D Large Language Models Really Understand 3D Spatial Relationships?

ReSpace: Text-Driven 3D Scene Synthesis and Editing with Preference Alignment

SGAligner++: Cross-Modal Language-Aided 3D Scene Graph Alignment

Rectified Point Flow: Generic Point Cloud Pose Estimation

GuideFlow3D: Optimization-guided Rectified Flow for Appearance Transfer

Facade Segmentation for Solar Photovoltaic Suitability

HouseTour: A Virtual Real Estate A(I)gent

ReStyle3D: Scene-level Appearance Transfer with Semantic Correspondences

WildGS-SLAM: Monocular Gaussian Splatting SLAM in Dynamic Environments

CrossOver: Scene Cross-Modal Alignment

LoopSplat: Loop Closure by Registering 3D Gaussian Splats

Multi-Hexplanes: A Lightweight Map Representation for Rendering and 3D Reconstruction

MAP-ADAPT: Real-Time Quality-Adaptive Semantic 3D Maps

"Where am I?" Scene Retrieval with Language

I-Design: Personalized LLM Interior Designer

Nothing Stands Still: A Spatiotemporal Benchmark on 3D Point Cloud Registration Under Large Geometric and Temporal Change

Living Scenes: Multi-object Relocalization and Reconstruction in Changing 3D Environments

Multiway Point Cloud Mosaicking with Diffusion and Global Optimization

Semantically Guided Feature Matching for Visual SLAM

Volumetric Semantically Consistent 3D Panoptic Mapping

Q-REG: End-to-End Trainable Point Cloud Registration with Surface Curvature

SGAligner: 3D Scene Alignment with Scene Graphs

ARrow: A Real-Time AR Rowing Coach

Learning-Based Relational Object Matching Across Views

HoloLabel: Augmented Reality User-In-The-Loop Online Annotation Tool for As-Is Building Information

SemSpray: Virtual Reality As-Is Semantic Information Labeling Tool for 3D Spatial Data

ImpliCity: City Modeling From Satellite Images with Deep Implicit Occupancy Fields

Robust Policies via Mid-Level Visual Representations: An Experimental Study in Manipulation and Navigation

3D Scene Graph: A Structure for Unified Semantics, 3D Space, and Camera

SEGCloud: Semantic Segmentation of 3D Point Clouds

Joint 2D-3D-Semantic Data for Indoor Scene Understanding

3D Semantic Parsing of Large-Scale Indoor Spaces

State of Research in Automatic As-Built Modelling

A dynamic identification of a historical building using accelerometers with interface modules and a digital synchronization method

Pedestrian navigation and shortest path: Preference versus distance

Resume of the thesis "More than a machine" J. A. Coderch

Books and chapters

Artificial Intelligence for Predicting Reuse Patterns

Selected conversations and stories

Panel Discussion on Practical Applications of AI: On Construction

Computer Vision and Mixed Reality in Architecture with Prof. Iro Armeni

A Day in the Life of an Architect in the Gradient World

Open data and resources

Open to collaborations, research, and conversations.