Learning Basis Representation to Refine 3D Human Pose Estimations

Authors: Chunyu Wang, Haibo Qiu, Alan L. Yuille, Wenjun Zeng8925-8932

AAAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on benchmark datasets show that our approach obtains more legitimate poses over the baselines.
Researcher Affiliation Collaboration Microsoft Research Asia, Beijing, China The Johns Hopkins University, Baltimore, MD 21218, USA
Pseudocode No No pseudocode or algorithm blocks are present. The methodology is described through text and mathematical formulations.
Open Source Code No No explicit statement about releasing their own source code or a link to it is provided.
Open Datasets Yes We evaluate our 3D pose refinement approach on two benchmark datasets: H36M (Ionescu et al. 2014) and MPIINF-3DHP (Mehta et al. 2017).
Dataset Splits No Following the most common evaluation protocol (Zhou et al. 2017; Pavlakos et al. 2017), we use five subjects (i.e. S1, S5, S6, S7, S8) for training and two subjects (S9, S11) for testing.
Hardware Specification No No specific hardware details (e.g., GPU/CPU models, memory) used for running the experiments are provided.
Software Dependencies No Only “Pytorch” is mentioned, but no specific version number or other software dependencies with their versions are listed.
Experiment Setup Yes We learn 1000 bases for all the training poses.