Hi there 👋

• 🍿This repo lists the generative AI works at IDEA.
• 👩‍💻Stay tuned for timely updates.

🔥News

[PEAR] Our latest human mesh recovery paper is accpeted in Siggraph 2026. It is an ultra fast at 100FPS and accurate motion capture tool. Code.

Talking Head / Face Editing / Avatar Generation

• [MODA] Mapping-Once Audio-driven Portrait Animation with Dual Attentions, ICCV 2023. Code.
• [GOAE] Make Encoder Great Again in 3D GAN Inversion through Geometry and Occlusion-Aware Encoding, ICCV 2023.
• [GPAvatar] GPAvatar: Generalizable and Precise Head Avatar from Image(s), ICLR 2024.
• [TEASER] TEASER: Token Enhanced Spatial Modeling for Expressions Reconstruction, ICLR 2025.
• [AnyTalk] AnyTalk: Multi-modal Driven Multi-domain Talking Head Generation, AAAI 2025.

2D Generation / Editing

• [HumanSD] HumanSD: A Native Skeleton-Guided Diffusion Model for Human Image Generation, ICCV 2023.
• [Direct Inversion] Boosting Diffusion-based Editing with 3 Lines of Code, ICLR 2024.
• [AddMe] AddMe: Zero-Shot Group-Photo Synthesis by Inserting People Into Scenes, ECCV 2024.

Video Generation / Editing

• [STEM] A Video is Worth 256 Bases: Spatial-Temporal Expectation-Maximization Inversion for Zero-Shot Video Editing, CVPR 2024.

Motion Generation

• [DiffSHEG] A Diffusion-Based Approach for Real-Time Speech-driven Holistic 3D Expression and Gesture Generation, CVPR 2024.
• [HumanTOMATO] Text-aligned Whole-body Motion Generation, ICML 2024.

3D Generation

• [DreamWaltz] Make a Scene with Complex 3D Animatable Avatars, NeurIPS 2023.
• [DreamTime] An Improved Optimization Strategy for Text-to-3D Content Creation, ICLR 2024.
• [Progressive3D] Progressively Local Editing for Text-to-3D Content Creation with Complex Semantic Prompts, ICLR 2024.
• [TOSS] High-quality Text-guided Novel View Synthesis from a Single Image, ICLR 2024.
• [Compress3D] A Compressed Latent Space for 3D Generation from a Single Image, ECCV 2024.
• [Consistent123] Improve Consistency for One Image to 3D Object Synthesis, Arxiv.
• [SSD] Stable Score Distillation for High-Quality 3D Generation, Arxiv.

Motion Capture

• [EDPose] Explicit Box Detection Unifies End-to-End Multi-Person Pose Estimation, ICLR 2023.
• [ClickPose] Neural Interactive Keypoint Detection, ICCV 2023.
• [OSX] One-Stage 3D Whole-Body Mesh Recovery with Component Aware Transformer, CVPR 2023.
• [DWPose] Effective Whole-body Pose Estimation with Two-stages Distillation, ICCV-W 2023.
• [AiOS] AiOS: All-in-One-Stage 3D Wholebody Mesh Recovery, CVPR 2024.

Datasets

• [Human-Art] Human-Art: A Versatile Human-Centric Dataset Bridging Natural and Artificial Scenes CVPR 2023.
• [GHI & Laion-Human] HumanSD: A Native Skeleton-Guided Diffusion Model for Human Image Generation, ICCV 2023.
• [UBody] One-Stage 3D Whole-Body Mesh Recovery with Component Aware Transformer, CVPR 2023.
• [Motion-X] Motion-X: A Large-scale 3D Expressive Whole-body Human Motion Dataset, NeurIPS 2023 Datasets and Benchmarks Track.
• [FreeMan] FreeMan: Towards benchmarking 3D human pose estimation under Real-World Conditions, CVPR 2024.

Join Us

Explore our current intern opportunities by visiting link1, link2, link3.