ECCV 2026: EquiFusion Enables Kinematics-Agnostic Human Motion Prediction
CSProfKGD · x · 2026-09-01
Researchers from TUM and MCML introduce EquiFusion, the first kinematics-agnostic stochastic human motion prediction model accepted at ECCV 2026. It utilizes a permutation-equivariant latent diffusion architecture that treats skeletal connectivity as an explicit input parameter, making computations inherently agnostic to joint ordering. This allows for true cross-dataset generalization to unseen kinematics and enables zero-shot capabilities like predicting from occluded observations. EquiFusion achieves state-of-the-art results on major benchmarks (AMASS, H36M, Nymeria), is up to 75% more compact than previous methods, and offers faster training and inference.
More from Research
- AI Research Preference Models: Using Compute to Decide What Experiments to Run — anirudhg9119 · 2026-09-01
- RPM-guided search sets new SOTA on benchmarks, proving selection scales with compute — anirudhg9119 · 2026-09-01
- Inference-only RPM: Frozen LLM Ranks Candidates to Save Compute — anirudhg9119 · 2026-09-01
- From score forecasting to preference ranking: picking which of 15 ideas to run — anirudhg9119 · 2026-09-01
- AI Research Agents Face Bottleneck: Ideas Outpace GPU Budget — anirudhg9119 · 2026-09-01
- Paper: AI Research Preference Models Solve Compute Bottleneck — anirudhg9119 · 2026-09-01