Existence Is Chaos: Enhancing 3D Human Motion Prediction with Uncertainty Consideration
Authors: Zhihao Wang, Yulin Zhou, Ningyu Zhang, Xiaosong Yang, Jun Xiao, Zhao Wang
AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | To evaluate our idea, extensive experiments are carried out on H3.6M, 3DPW and CMU Mocap datasets to study the impact of learning early frames for the final performance. The results demonstrate that our method achieve competitive performance in both short-term and long-term motion prediction tasks. |
| Researcher Affiliation | Academia | 1Zhejiang University 2Ningbo Innovation Center, Zhejiang University 3National Centre for Computer Animation, Bournemouth University |
| Pseudocode | No | The paper does not contain structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | The code will be available in https://github.com/Motionpre/Adaptive-Salient-Loss-SAGGB. |
| Open Datasets | Yes | To evaluate our idea, extensive experiments are carried out on H3.6M, 3DPW and CMU Mocap datasets to study the impact of learning early frames for the final performance. |
| Dataset Splits | Yes | 3D Pose in the Wild dataset (3DPW) includes both indoor and outdoor actions captured at 30Hz. Each pose has 26 joints and we use 23 of them. We conduct experiments according to the official segmented training set, validation set and test set. |
| Hardware Specification | No | The paper does not provide specific hardware details (exact GPU/CPU models, processor types, or memory amounts) used for running its experiments. |
| Software Dependencies | No | The paper does not provide specific ancillary software details with version numbers (e.g., library or solver names with version numbers like Python 3.8, CPLEX 12.4) needed to replicate the experiment. |
| Experiment Setup | Yes | The ablation study results for hyper parameters is shown in Fig. 5. These results confirm the validity of our proposed loss and also support our hypothesis. The salient factor ω and weight factor λ is determined as 10 and 0.3, respectively. |