I’m passionate about advancing Spatial Intelligence.

• SpatialGen (3DV 2026): Multi-view multi-modal diffusion model for 3D indoor scene generation.
• SpatialLM (NeurIPS 2025): Large Language Model for structured indoor scene modeling.
• NonCuboidRoom (WACV 2022): Single-view 3D room layout reconstruction beyond cuboid assumptions.
• Structured3D (ECCV 2020): Large-scale, photo-realistic dataset for structured 3D scene reconstruction.
• PlaneAE (CVPR 2019): Piece-wise planar 3D reconstruction from a single RGB image.

• CAD2Program (AAAI 2025): Vision-language model for reconstructing 3D parametric CAD models from 2D drawings.
• PlankAssembly (ICCV 2023): First deep generative model for 3D CAD reconstruction based on three orthographic views.
• Faceformer (CVPR 2022): Data-driven approach to identify face loops from single line drawing.

• FLUX.1-Panorama-LoRA: FLUX.1 LoRA for generating high-quality panoramic images focusing on realistic interior design scenes.
• FLUX.1-Wireframe-dev-LoRA: FLUX.1 LoRA for generating images based on a text description and a wireframe input.
• KoolCogVideoX: Fine-tuned CogVideoX-based text-to-video model for enhanced performance in generating realistic interior design videos.

• awesome-CAD: Curated papers on Computer-Aided Design (CAD).
• awesome-scene-understanding: Curated papers on scene understanding.
• awesome-holistic-3d: Curated papers on holistic 3D reconstruction.