Inwoo Hwang

I'm a PhD student in 3D Vision lab at Seoul National University, advised by Prof. Young Min Kim. I was a research intern at Snap Research.

My research focuses on building practical motion generation systems that enable controllable and robust human motion modeling. Recently, I have been focusing on egocentric motion reconstruction and text-to-motion generation.

/ CV / LinkedIn / Google Scholar

Publications

Research Keywords: Character Animation, Human Interaction, Controllable Motion Synthesis, Robust Motion Reconstruction.

	SnapMoGen: Human Motion Generation from Expressive Texts Chuan Guo, Inwoo Hwang, Jian Wang, and Bing Zhou NeurIPS, 2025 arxiv \| project page Large-scale text-motion dataset featuring high-quality motion capture data paired with accurate, expressive textual annotations.
	SceneMI: Motion In-betweening for Modeling Human-Scene Interaction Inwoo Hwang, Bing Zhou, Young Min Kim, Jian Wang, and Chuan Guo ICCV, 2025 (Highlight) arxiv \| project page Modeling Human-Scene Interaction (HSI) as scene-aware motion in-betweening, and supports various practical applications including video-based HSI reconstruction.
	Motion Synthesis with Sparse and Flexible Keyjoint Control Inwoo Hwang, Jinseok Bae, Donggeun Lim, and Young Min Kim ICCV, 2025 arxiv \| project page A controllable motion synthesis pipeline with high quality and precision, from explicit or implicit control signals, including time-agnostic motion control.
	Less is More: Improving Motion Diffusion Models with Sparse Keyframes Jinseok Bae, Inwoo Hwang, Young Yoon Lee, Ziyu Guo, Joseph Liu, Yizhak Ben-Shabat, Young Min Kim, and Mubbasir Kapadia ICCV, 2025 arxiv \| project page A sparse keyframe-based motion diffusion model that better captures text prompts and improves overall motion quality.
	Event-Driven Storytelling with Multiple Lifelike Humans in a 3D Scene Donggeun Lim, Jinseok Bae, Inwoo Hwang, Seungmin Lee, Hwanhee Lee, and Young Min Kim ICCV, 2025 arxiv \| project page A framework that creates a lively virtual dynamic scene with contextual motions of multiple humans.
	A Survey on Human Interaction Motion Generation Kewei Sui, Anindita Ghosh, Inwoo Hwang, Bing Zhou, Jian Wang, and Chuan Guo IJCV, 2025 arxiv \| project page A review of recent advances in human interaction motion generation, including human-human, human-object, human-scene, and human-mix interactions.
	Goal-Driven Human Motion Synthesis in Diverse Tasks Inwoo Hwang, Jinseok Bae, Donggeun Lim, and Young Min Kim CVPRW, 2025, HuMoGen paper \| project page A motion generation pipeline from predefined key joint goal positions and a 3D environment.
	Versatile Physics-based Character Control with Hybrid Latent Representation Jinseok Bae, Jungdam Won, Donggeun Lim, Inwoo Hwang, and Young Min Kim Eurographics, 2025 arxiv \| paper \| project page Propose integrating continuous and discrete latent representations, enabling physically simulated characters to efficiently utilize motion priors and adapt to diverse challenging control tasks.
	Text2Scene: Text-driven Indoor Scene Stylization with Part-aware Details Inwoo Hwang, Hyeonwoo Kim, and Young Min Kim CVPR, 2023 (Highlight) arxiv \| paper \| video \| project page A method to automatically create realistic and part-aware textures for virtual scenes composed of multiple objects.
	Text2PointCloud: Text-Driven Stylization for Sparse PointCloud Inwoo Hwang, Hyeonwoo Kim, Donggeun Lim, Inbum Park, and Young Min Kim Eurographics short, 2023 paper \| video A method to process sparse, noisy point cloud input and generate high-quality stylized output.
	Ev-NeRF: Event Based Neural Radiance Field Inwoo Hwang, Junho Kim, and Young Min Kim WACV, 2023 arxiv \| paper \| video \| project page A Neural Radiance Field (NeRF) derived from event data, which serves as solutions for various event-based applications and highly robust to sensor noise.
	Ev-TTA: Test-Time Adaptation for Event-Based Object Recognition Junho Kim, Inwoo Hwang, and Young Min Kim CVPR, 2022 arxiv \| paper \| video \| project page A simple and effective test-time adaptation algorithms for event-based object recognition. Successfully adapt classifiers to various external conditions.
	MasKGrasp: Mask-based Grasping for Scenes with Multiple General Real-world Objects Junho Lee, Junhwa Hur, Inwoo Hwang, and Young Min Kim IROS, 2022 paper \| video A real world grasping algorithm that can generalize to transparent and opaque obejcts via masks.