Research map / Video / Video generation
SPARD: Single-step Inference with Adaptive Sampling in Residual Diffusion for Human Motion Prediction
Yiming Zhang, Baojia Han, Ximing Li, Wei Pang, Fausto Giunchiglia, Xiaoyue Feng, Renchu Guan
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 noise predictor module (ablation) | SPARD (distilled) · Human3.6M | ADE: 0.342 -> 0.318 | Improved |
| add noise predictor module (ablation) | SPARD (distilled) · Human3.6M | APD: 6.338 -> 5.571 | Hurt |
| add distillation to single-step student (ablation) | SPARD · Human3.6M | inference time (ms): 24.72 -> 2.52 | Improved |
| add distillation to single-step student (ablation) | SPARD · Human3.6M | ADE / APD: ADE 0.336 -> 0.342; APD 7.500 -> 6.338 | Mixed |
| increase relaxation predictions h from 1 to 10 (ablation) | SPARD residual diffusion · Human3.6M | APD: 3.192 -> 7.500 | Improved |
| increase relaxation predictions h from 1 to 10 (ablation) | SPARD residual diffusion · Human3.6M | training time per epoch (s): 39.98 -> 113.49 | Hurt |
| increase guidance coefficient s from 0.1 to 0.3 (ablation) | SPARD residual diffusion · Human3.6M | ADE: 0.344 -> 0.336 | Improved |
| increase guidance coefficient s to 0.7 (ablation) | SPARD residual diffusion · Human3.6M | MMFDE: 0.499 -> 0.526 | Hurt |
| single-step SPARD vs fastest diffusion baseline BeLFusion | SPARD · Human3.6M | inference time (ms): 45.03 -> 2.82 | 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.
- BeLFusion — BeLFusion: Latent Diffusion for Behavior-Driven Human Motion Prediction (ICCV 2023) · better on 5 of 6 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
- WaveAR: Wavelet-Aware Continuous Autoregressive Diffusion for Accurate Human Motion Prediction (NeurIPS 2025 · similarity 0.90)
- MotionMix: Weakly-Supervised Diffusion for Controllable Motion Generation (AAAI 2024 · similarity 0.88)
- Human Motion Diffusion as a Generative Prior (ICLR 2024 · similarity 0.88)
- SingularTrajectory: Universal Trajectory Predictor Using Diffusion Model (CVPR 2024 · similarity 0.88)
- Optimizing Diffusion Noise Can Serve As Universal Motion Priors (CVPR 2024 · similarity 0.88)
- Rethinking Diffusion for Text-Driven Human Motion Generation: Redundant Representations, Evaluation, and Masked Autoregression (CVPR 2025 · similarity 0.87)
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.
