HybridCap: Inertia-Aid Monocular Capture of Challenging Human Motions
Authors: Han Liang, Yannan He, Chengfeng Zhao, Mutian Li, Jingya Wang, Jingyi Yu, Lan Xu
AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments on various datasets demonstrate Hybrid Cap can robustly handle challenging movements ranging from fitness actions to Latin dance. It also achieves real-time performance up to 60 fps with state-of-the-art accuracy. |
| Researcher Affiliation | Academia | 1School of Information Science and Technology, Shanghai Tech University 2Shanghai Frontiers Science Center of Human-centered Artificial Intelligence |
| Pseudocode | No | The paper does not contain structured pseudocode or algorithm blocks, nor does it refer to any explicitly labeled "Algorithm" section. |
| Open Source Code | No | Our code and HCM dataset will be made publicly available to stimulate future research. |
| Open Datasets | Yes | To provide sufficient 2D and inertial observations, we build and propose a new dataset Hybrid Challenging Motions (HCM), which consists of abundant records of RGB cameras and IMUs. Please refer to the supplementary to appreciate more details. We further utilize AIST++ (Li et al. 2021; Tsuchida et al. 2019) which consists of various challenging dance sequences. Our code and HCM dataset will be made publicly available to stimulate future research. |
| Dataset Splits | No | The paper states, "We follow the standard protocols to split out test sets for all the datasets," but does not explicitly provide specific details about training, validation, and test splits (percentages or counts) that would be needed for reproduction. |
| Hardware Specification | Yes | We run our pipeline on a PC with an i7-10700k CPU and RTX3070 GPU, where the inference module and optimization module take 2.8( 0.4) ms and 6.2( 3.5) ms respectively, achieving 60 fps. |
| Software Dependencies | No | The paper mentions software like Open Pose and Ceres Solver, but does not provide specific version numbers for these or other key software dependencies required for replication. |
| Experiment Setup | Yes | In our experiments, we empirically set the parameters λ3D = 10, λ2D = 1, λacc = 10 and λori = 30. |