Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
ToF-IP: Time-of-Flight Enhanced Sparse Inertial Poser for Real-time Human Motion Capture
Authors: Yuan Yao, Shifan Jiang, Yangqing Hou, Chengxu Zuo, Xinrui Chen, Shihui Guo, Yipeng Qin
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments demonstrate that our method outperforms state-of-the-art approaches such as PNP, achieving superior accuracy in tracking complex and slow motions like Tai Chi, which remains challenging for inertial-only methods. ... We contribute a 208-minute human motion dataset from 10 participants, including synchronized IMU-To F measurements and groundtruth from optical tracking. ... Extensive experimental results show that, compared to state-of-the-art (SOTA) methods, our approach significantly improves joint position estimation, achieving superior accuracy in tracking complex and slow movements like Tai Chi. |
| Researcher Affiliation | Academia | 1 School of Informatics, Xiamen University, China 2 School of Computer Science & Informatics, Cardiff University, UK |
| Pseudocode | No | The paper describes methods like "Node-Centric Data Integration" and "Dynamic Spatial Positional Encoding" using mathematical equations (Eq. 1-5) and prose, but does not include a clearly labeled pseudocode or algorithm block. |
| Open Source Code | No | Question: Does the paper provide open access to the data and code, with sufficient instructions to faithfully reproduce the main experimental results, as described in supplemental material? Answer: [Yes] Justification: Yes, we will release the code/data later. |
| Open Datasets | No | We propose To F-IP-DB, a large dataset containing over 20 types of motion activities, 208 minutes (749,000 frames) collected from 10 participants (3 male, 7 female), including dynamic motions such as dances and aerobics, as well as slow-paced movements like Tai Chi and Baduanjin. This dataset uniquely combines synchronized To F distance maps, 6-Do F IMU signals, and SMPL reference poses, with GT motion data. ... Question: Does the paper provide open access to the data and code, with sufficient instructions to faithfully reproduce the main experimental results, as described in supplemental material? Answer: [Yes] Justification: Yes, we will release the code/data later. |
| Dataset Splits | No | The paper mentions using the AMASS dataset for pre-training and introducing their own To F-IP-DB dataset. It provides the total size of their collected dataset (208 minutes, 749,000 frames from 10 participants) but does not specify the train/test/validation splits for either dataset in the main text, nor does it provide exact percentages or sample counts for how the data was partitioned. |
| Hardware Specification | Yes | All our experiments run on a PC with an Intel(R) Core(TM) i7-13700KF CPU and an NVIDIA RTX 4080 GPU. |
| Software Dependencies | Yes | The model is implemented using Py Torch 1.12.1 with CUDA 11.3. |
| Experiment Setup | Yes | We use the Adam optimizer with a learning rate of lr = 1 10 3 and weight decay of lr = 1 10 6 during n epochs training. The batch size was set to 512. |