Zihan Zhou
Chief Scientist · Manycore Tech Inc.
Hangzhou, Zhejiang, P. R. China
I am currently the Chief Scientist at Manycore Tech Inc. From 2013 to 2021, I was a research associate and then an assistant professor in the College of Information Sciences and Technology at Penn State University. I completed my Ph.D. in Electrical and Computer Engineering at the University of Illinois at Urbana-Champaign in 2013, advised by Prof. Yi Ma. I received my Bachelor degree in Automation from Tsinghua University and my Master degree in ECE from UIUC.
People's Daily: The "Toolmakers" of Spatial Intelligence
A People's Daily feature on how Manycore is building the foundational infrastructure powering spatial intelligence and embodied AI.
SpatialLM: Teaching LLMs to Understand 3D Space
A feature on our open-source spatial language model that reads structured room layouts directly from point-cloud data.
World Models for Physical Space Intelligence
Building world models that perceive, understand, and generate physical 3D space — teaching machines to read structured layouts from raw sensor data and to synthesize coherent, photorealistic indoor environments.
SpatialGen: Layout-guided 3D Indoor Scene Generation
3DV 2026
SpatialLM: Training Large Language Models for Structured Indoor Modeling
NeurIPS 2025
Learning 3D Structures
Recovering the structural regularities of man-made environments — planes, wireframes, and superpixels — from single images, and building large-scale datasets to learn geometry-aware 3D representations.
Structured3D: A Large Photo-realistic Dataset for Structured 3D Modeling
ECCV 2020
Superpixel Segmentation with Fully Convolutional Networks
CVPR 2020
Single-Image Piece-Wise Planar 3D Reconstruction via Associative Embedding
CVPR 2019
Recovering 3D Planes from a Single Image via Convolutional Neural Networks
ECCV 2018
Learning to Parse Wireframes in Images of Man-Made Environments
CVPR 2018