Expanding mmWave Datasets for Human Pose Estimation with Unlabeled Data and LiDAR Datasets
This addresses a domain-specific problem for researchers in human pose estimation by improving dataset availability and model generalization, though it is incremental as it builds on existing data augmentation techniques.
The paper tackled the scarcity and lack of diversity in mmWave datasets for human pose estimation by proposing EMDUL, which expands datasets using unlabeled mmWave data and LiDAR datasets, resulting in error reductions of 15.1% and 18.9% for in-domain and out-of-domain settings.
Current mmWave datasets for human pose estimation (HPE) are scarce and lack diversity in both point cloud (PC) attributes and human poses, severely hampering the generalization ability of their trained models. On the other hand, unlabeled mmWave HPE data and diverse LiDAR HPE datasets are readily available. We propose EMDUL, a novel approach to expand the volume and diversity of an existing mmWave dataset using unlabeled mmWave data and a LiDAR dataset. EMDUL trains a pseudo-label estimator to annotate the unlabeled mmWave data and is able to convert, or translate, a given annotated LiDAR PC to its mmWave counterpart. Expanded with both LiDAR-converted and pseudo-labeled mmWave PCs, our mmWave dataset significantly boosts the performance and generalization ability of all our HPE models, with substantial 15.1% and 18.9% error reductions for in-domain and out-of-domain settings, respectively.