Research map / Video / Video generation
BeLFusion: Latent Diffusion for Behavior-Driven Human Motion Prediction
German Barquero, Sergio Escalera, Cristina Palmero
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 behavioral latent space (BLS) to the latent loss and reconstruction loss (ablation) | BeLFusion · Human3.6M | ADE: 0.380 -> 0.372 | Improved |
| add latent loss L_lat on top of reconstruction loss (ablation) | BeLFusion · Human3.6M | ADE: 0.386 -> 0.372 | Improved |
| train with reconstruction loss only (no latent loss, no BLS) (ablation) | BeLFusion · AMASS cross-dataset | ADE: 0.513 -> 0.697 | Hurt |
| run all denoising steps instead of a single step (ablation) | BeLFusion · Human3.6M | CMD: 8.508 -> 5.988 | Improved |
| run all denoising steps instead of a single step (ablation) | BeLFusion · Human3.6M | ADE: 0.367 -> 0.372 | Mixed |
| increase relaxation parameter k (ablation) | BeLFusion · Human3.6M | diversity, accuracy and realism | Improved |
| use BeLFusion instead of GSPS and DivSamp | BeLFusion · AMASS cross-dataset | ADE: 0.564 -> 0.513 | Improved |
| use BeLFusion instead of DivSamp for diversity | BeLFusion · Human3.6M | APD: 15.310 -> 7.602 | Hurt |
| use BeLFusion in human realism study | BeLFusion · Human3.6M user study | average realism rank: 2.246 -> 1.415 | Improved |
Later papers that compared with this method
3 later accepted papers report BeLFusion as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- SPARD: Single-step Inference with Adaptive Sampling in Residual Diffusion for Human Motion Prediction (AAAI 2026 · better on 5 of 6 shared results)
- Learning Semantic Latent Directions for Accurate and Controllable Human Motion Prediction (ECCV 2024 · better on 2 of 2 shared results)
- HumanMAC: Masked Motion Completion for Human Motion Prediction (ICCV 2023 · better on 2 of 3 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
- CoMusion: Towards Consistent Stochastic Human Motion Prediction via Motion Diffusion (ECCV 2024 · similarity 0.90)
- Human Joint Kinematics Diffusion-Refinement for Stochastic Motion Prediction (AAAI 2023 · similarity 0.90)
- Nonisotropic Gaussian Diffusion for Realistic 3D Human Motion Prediction (CVPR 2025 · similarity 0.89)
- FutureHuman3D: Forecasting Complex Long-Term 3D Human Behavior from Video Observations (CVPR 2024 · similarity 0.88)
- Human Motion Diffusion as a Generative Prior (ICLR 2024 · similarity 0.88)
- Motion Diversification Networks (CVPR 2024 · similarity 0.88)
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.
