Research map / 3D vision / Human pose estimation
Progressive Guessing to Fixed Point: Rethinking Human Motion Prediction with Deep Equilibrium Models
Dong Wei, Huaijiang Sun, Fan Liu, Yuhui Zheng
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 |
|---|---|---|---|
| use last observed frame as initial guess instead of observed motion (ablation) | DEQ motion predictor · Human3.6M 400ms | MPJPE: 104.6 -> 55.3 | Improved |
| add equivariant geometric injection (ablation) | MotionDEQ · Human3.6M 400ms | MPJPE: 64.0 -> 55.3 | Improved |
| add invariant pattern injection (ablation) | MotionDEQ · Human3.6M 400ms | MPJPE: 61.3 -> 55.3 | Improved |
| use sparse fixed-point supervision instead of smoothed GT targets (ablation) | MotionDEQ · 3DPW 1000ms | MPJPE: 120.59 -> 107.80 | Improved |
| use sparse fixed-point supervision instead of dense corrections (ablation) | MotionDEQ · 3DPW 1000ms | MPJPE: 114.09 -> 107.80 | Improved |
| use truncated (phantom) gradient instead of inexact JFB gradient (ablation) | MotionDEQ · Human3.6M | MPJPE | Improved |
| increase DEQ forward iterations up to about 10 (ablation) | MotionDEQ · Human3.6M | MPJPE | Improved |
| share parameters across refinement stages (RNN-origin) without last-frame init (ablation) | EqMotion · 3DPW | Avg MPJPE: 67.07 -> 69.22 | Hurt |
| replace multi-stage EqMotion with MotionDEQ | EqMotion · Human3.6M 400ms | MPJPE: 55.9 -> 55.3 | Improved |
| replace multi-stage EqMotion with MotionDEQ | EqMotion · 3DPW average | MPJPE: 67.07 -> 65.95 | Improved |
| reuse previous fixed point as warm initial guess | MotionDEQ · streaming Human3.6M | DEQ iterations to converge | Improved |
| add equivariant correction adapter (ECA) | MotionDEQ · streaming Human3.6M | MPJPE | Mixed |
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.
- ALIEN — ALIEN: Implicit Neural Representations for Human Motion Prediction under Arbitrary Latency (CVPR 2025) · better on 1 of 1 shared result
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
- Learning Progressive Joint Propagation for Human Motion Prediction (ECCV 2020 · similarity 0.88)
- Context-Aware Human Motion Prediction (CVPR 2020 · similarity 0.88)
- Learning Semantic Latent Directions for Accurate and Controllable Human Motion Prediction (ECCV 2024 · similarity 0.88)
- Adversarial Geometry-Aware Human Motion Prediction (ECCV 2018 · similarity 0.88)
- Human Joint Kinematics Diffusion-Refinement for Stochastic Motion Prediction (AAAI 2023 · similarity 0.88)
- Motion Diversification Networks (CVPR 2024 · 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.
