Research map / Video / Video generation
CrowdMoGen: Event-Driven Collective Human Motion Generation
Yukang Cao, Xinying Guo, Mingyuan Zhang, Haozhe Xie, Chenyang Gu, Ziwei Liu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add ControlAttention (ablation) | CrowdMoGen · HumanML3D ablation | FID: 0.150 -> 0.127 | Improved |
| add InputMixing (ablation) | CrowdMoGen · HumanML3D ablation | FID: 0.207 -> 0.127 | Improved |
| add controlled keypoint loss L_con (ablation) | CrowdMoGen · HumanML3D ablation | FID: 0.219 -> 0.127 | Improved |
| add foot skating loss L_foot (ablation) | CrowdMoGen · HumanML3D ablation | Foot skating ratio: 0.1947 -> 0.0762 | Improved |
| add foot skating loss L_foot (ablation) | CrowdMoGen · HumanML3D ablation | Avg. error (m): 0.0147 -> 0.0196 | Hurt |
| add IK guidance at inference (ablation) | CrowdMoGen · HumanML3D ablation | Avg. error (m): 0.2387 -> 0.0196 | Improved |
| add IK guidance at inference (ablation) | CrowdMoGen · HumanML3D ablation | FID: 0.130 -> 0.127 | No clear effect |
| replace ControlNet-only control with InputMixing + ControlAttention (ablation) | CrowdMoGen · HumanML3D ablation | FID: 0.225 -> 0.127 | Improved |
| replace InterControl with CrowdMoGen | text-to-motion · HumanML3D controllable (Pelvis) | FID: 0.159 -> 0.132 | Improved |
| replace InterControl with CrowdMoGen | text-to-motion · KIT-ML text-to-motion | FID: 0.580 -> 0.217 | Improved |
| use CrowdMoGen vs OmniControl on foot skating | controllable generation · HumanML3D controllable (Pelvis) | Foot skating ratio: 0.0547 -> 0.0762 | Hurt |
| use full CrowdMoGen vs w/o ControlAttention in user study | CrowdMoGen · User study (25 volunteers) | Motion quality preference: 13.04% -> 77.91% | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- InterControl — InterControl: Zero-shot Human Interaction Generation by Controlling Every Joint (NeurIPS 2024) · better on 9 of 9 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Human Motion Diffusion as a Generative Prior (ICLR 2024 · similarity 0.87)
- AMD: Autoregressive Motion Diffusion (AAAI 2024 · similarity 0.86)
- SnapMoGen: Human Motion Generation from Expressive Texts (NeurIPS 2025 · similarity 0.86)
- Continuous Locomotive Crowd Behavior Generation (CVPR 2025 · similarity 0.86)
- Learning to Generate Diverse Pedestrian Movements from Web Videos with Noisy Labels (ICLR 2025 · similarity 0.86)
- Motion-Agent: A Conversational Framework for Human Motion Generation with LLMs (ICLR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
