
Debao Huang
I’m a Research Scientist at Spectral Labs, building spatial intelligence foundation models for engineering physical systems.
I received my Ph.D. in Electrical and Computer Engineering from The Ohio State University in 2026, advised by Prof. Rongjun Qin in the Geospatial Data Analytics Group. My research focuses on large-scale 3D reconstruction (SfM, NeRF, 3DGS, 4DGS) and uncertainty quantification. Additionally, I leverage vision foundation models and multimodal learning to advance large-scale 3D scene understanding. My doctoral research was supported by ONR, IARPA, AFRL, and NSF.
Keywords: 3D Computer Vision, Photogrammetry, Machine Learning, Generative AI, Vision Foundation Models.
News
10/2026 — Our team at Spectral Labs built the first AI system to pass a professional CAD certification. Using Quartz and SGS-Mk2, our agent passed the Certified SOLIDWORKS Professional (CSWP) exam, autonomously creating parts, modifying designs, and assembling mechanisms. Read A Turing Test for Physical Engineering and watch the announcement on LinkedIn or X.
08/2026 — I officially graduated with a Ph.D. in Electrical and Computer Engineering from The Ohio State University. Many thanks to my advisor, Prof. Rongjun Qin, and to my colleagues in the Geospatial Data Analytics Lab.
07/2026 — I successfully defended my Ph.D. dissertation, Uncertainty-Aware Photogrammetric Reconstruction from Multi-View Heterogeneous Data: Toward Accurate and Reliable 3D Data. Thanks to my advisor Prof. Rongjun Qin and my committee members, Prof. Alper Yilmaz and Prof. Han-Wei Shen.
05/2026 — I started a new role as a Research Scientist at Spectral Labs in the San Francisco Bay Area, building spatial intelligence foundation models for engineering physical systems.
05/2026 — Our paper, Uncertainty Quantification Framework for Aerial and UAV Photogrammetry through Error Propagation, has been accepted to the ISPRS Journal of Photogrammetry and Remote Sensing! We propose a self-calibrating framework that propagates errors through both the SfM and MVS stages to quantify per-point uncertainty in aerial and UAV photogrammetric point clouds. Check out the paper and code for more details.
02/2026 — Our paper, Olbedo: An Albedo and Shading Aerial Dataset for Large-Scale Outdoor Environments, has been accepted to CVPR 2026! We are releasing the first large-scale, real-world aerial dataset designed specifically for outdoor albedo-shading decomposition in the wild. Check out the project page and paper for more details.
01/2026 — I attended Geo Week 2026 in Denver, connecting with the geospatial community and industry professionals.
12/2025 — I am honored to announce that I have been awarded the Lyman J. Ladner Memorial Scholarship from ASPRS!
03/2025 — Our new papers on dynamic 3D Gaussian Splatting and uncertainty quantification were accepted to ISPRS Geospatial Week 2025.
11/2024 — I was honored to be an outstanding reviewer for the 35th British Machine Vision Conference (BMVC 2024).
04/2024 — Paper on incorporating motion constraints in Bundle Adjustment published in ISPRS Journal of Photogrammetry and Remote Sensing.
04/2024 — Paper on scaling up NeRF for large-scale aerial images published in The Photogrammetric Record.
06/2023 — Paper on uncertainty quantification in MVS accepted to ISPRS Geospatial Week 2023.
05/2022 — Ohio State Engineering featured our PlanetScope research in Using satellite data to help direct response to natural disasters, highlighting satellite-based 3D reconstruction for disaster assessment.
05/2022 — Paper on baseline constraint in BA for SfM accepted to ISPRS Congress 2022.
04/2022 — Paper on PlanetScope images for 3D reconstruction published in GIScience & Remote Sensing Journal.
08/2021 — Paper on 3D change detection of building deconstruction published in International Journal of Digital Earth.
Selected Research
A Turing Test for Physical Engineering
Spectral Labs
Research release · Oct 14, 2026
Quartz and SGS-Mk2 combine CAD execution with geometric reasoning to build parts, edit designs, and assemble mechanisms. Our system is the first AI to pass a professional CAD certification. On exam problems involving geometric reasoning, SGS-Mk2 reduced per-problem costs by 46–87%.
Uncertainty Quantification Framework for Aerial and UAV Photogrammetry through Error Propagation
Debao Huang, Rongjun Qin
ISPRS Journal of Photogrammetry and Remote Sensing, 2026
This paper presents a novel, self-calibrating method to estimate the uncertainty in the MVS stage, and an uncertainty quantification framework for Aerial and UAV Photogrammetry.
Olbedo: An Albedo and Shading Aerial Dataset for Large-Scale Outdoor Environments
Shuang Song†, Debao Huang†, Deyan Deng, Haolin Xiong, Yang Tang, Yajie Zhao, Rongjun Qin
CVPR, 2026
Paper Project Page Code Demo Models Dataset
We introduce Olbedo, a large-scale aerial dataset containing 5,664 UAV images for outdoor albedo-shading decomposition in the wild.
Dynamic Urban Scene Modeling with 3D Gaussian Splatting from UAV Full Motion Videos
Debao Huang†, Hanyang Liu†, Ningli Xu, Rongjun Qin
ISPRS Archives, 2025
This work presents a method for reconstructing dynamic urban scenes from UAV full-motion videos using 3D Gaussian Splatting, effectively handling moving objects in large-scale environments.
Bundle Adjustment with Motion Constraints for Uncalibrated Multi-Camera Systems at the Ground Level
Debao Huang, Rongjun Qin, Mostafa Elhashash
ISPRS Journal of Photogrammetry and Remote Sensing, 2024
This paper derives novel geometric constraints in BA of SfM pipeline that assumes minimal knowns about the uncalibrated multi-camera systems at the ground level. Our proposed constraints are effective in correcting topographical errors (i.e., trajectory drifts) of the reconstruction.
Multi-Tiling Neural Radiance Field (NeRF)—Geometric Assessment on Large-Scale Aerial Datasets
Ningli Xu, Rongjun Qin, Debao Huang, Fabio Remondino
The Photogrammetric Record, 2024 (Cover article of 12/2024 issue)
This paper provides multi-camera tiling (MCT) strategy to scale the NeRF on large-scale aerial datasets and a thorough geometry assessment of NeRF.
An Evaluation of PlanetScope Images for 3D Reconstruction and Change Detection – Experimental Validations with Case Studies
Debao Huang, Yang Tang, Rongjun Qin
GIScience & Remote Sensing, 59(1), 744–761, 2022
We evaluate PlanetScope imagery for 3D surface reconstruction, changes in buildings and forests, and avalanche volume estimation.
Honors
• Lyman J. Ladner Memorial Scholarship, American Society for Photogrammetry and Remote Sensing, 2026 — awarded to outstanding students (Top 1%) pursuing research in photogrammetry.
Service
I serve as a reviewer for the following journals and conferences:
• International Journal of Computer Vision
• ISPRS Journal of Photogrammetry and Remote Sensing
• IEEE Transactions on Geoscience and Remote Sensing
• International Journal of Applied Earth Observation and Geoinformation
• Photogrammetric Engineering & Remote Sensing
• Remote Sensing
• Applied Sciences
• IEEE Transactions on Circuits and Systems for Video Technology
• International Journal of Digital Earth
• British Machine Vision Conference (BMVC 2024) — Outstanding Reviewer
• Conference on Neural Information Processing Systems (NeurIPS 2024)
© 2026 Debao Huang. All rights reserved.