Fast and scalable human pose estimation using mmWave point cloud

@article{An2022FastAS,
  title={Fast and scalable human pose estimation using mmWave point cloud},
  author={Sizhe An and Umit Y. Ogras},
  journal={Proceedings of the 59th ACM/IEEE Design Automation Conference},
  year={2022}
}
  • Sizhe An, U. Ogras
  • Published 29 April 2022
  • Computer Science
  • Proceedings of the 59th ACM/IEEE Design Automation Conference
Millimeter-Wave (mmWave) radar can enable high-resolution human pose estimation with low cost and computational requirements. However, mmWave data point cloud, the primary input to processing algorithms, is highly sparse and carries significantly less information than other alternatives such as video frames. Furthermore, the scarce labeled mmWave data impedes the development of machine learning (ML) models that can generalize to unseen scenarios. We propose a fast and scalable human pose… 

Figures and Tables from this paper

A Systematic Survey of Research Trends in Technology Usage for Parkinson’s Disease

TLDR
There is a substantial and steady growth in the use of mobileTechnology in the PD contexts, particularly in the last four years of the period under study, which reflects the research community's growing interest in assessing PD with wearable devices.

References

SHOWING 1-10 OF 25 REFERENCES

MARS: mmWave-based Assistive Rehabilitation System for Smart Healthcare

TLDR
This work proposes a millimeter-wave (mmWave)-based assistive rehabilitation system (MARS) for motor disorders that provides a low-cost solution with a competitive object localization and detection accuracy, and is the first rehabilitation movements dataset using mmWave point cloud.

mm-Pose: Real-Time Human Skeletal Posture Estimation Using mmWave Radars and CNNs

In this paper, mm-Pose, a novel approach to detect and track human skeletons in real-time using an mmWave radar, is proposed. To the best of the authors’ knowledge, this is the first method to detect

RF-based 3D skeletons

TLDR
RF-Pose3D is the first system that infers 3D human skeletons from RF signals, and works with multiple people and across walls and occlusions, and generates dynamic skeletons that follow the people as they move, walk or sit.

Learning Nondeterministic Real-Time Automata

TLDR
An active learning algorithm named NRTALearning for nondeterministic real-time automata (NRTAs) is presented and its correctness and polynomial complexity are proved, and it is proved that there exists a minimal RRTA to recognize any given real- time language.

mmMesh: towards 3D real-time dynamic human mesh construction using millimeter-wave

TLDR
The evaluation results on a commercial mmWave sensing testbed show that the proposed mmMesh system can accurately localize the vertices on the human mesh with an average error of 2.47 cm, which demonstrates the effectiveness of the proposed human mesh construction system.

Few-Shot Meta-Learning on Point Cloud for Semantic Segmentation

TLDR
This work pioneered the method of producing datasets for meta-learning in 3D scenes, as well as demonstrated that the Model-Agnostic Meta-Learning (MAML) algorithm can be applied to process 3D point cloud data.

Transfer Learning for Human Activity Recognition using Representational Analysis of Neural Networks

TLDR
This work proposes a HAR transfer learning framework that transfers the reusable portion of the offline classifier to new users and fine-tune only the rest, and shows up to 43% accuracy improvement and 66% training time reduction when compared to the baseline without using transfer learning.

Few Shot Learning For Point Cloud Data Using Model Agnostic Meta Learning

TLDR
The MAML algorithm is extended to point cloud data using a PointNet Architecture and shows that this method performs classification as well as supervised deep learning methods with the added benefit of being able to adapt after a single gradient step on a single N$\times$K task.

Rethinking Motion Representation: Residual Frames With 3D ConvNets

TLDR
The proposed residual-input path can outperform RGB counterpart on unseen datasets when the authors apply trained models to video retrieval tasks, and huge improvements can be obtained when the residual inputs are applied to video-based self-supervised learning methods.

A Comprehensive Survey on Transfer Learning

TLDR
This survey attempts to connect and systematize the existing transfer learning research studies, as well as to summarize and interpret the mechanisms and the strategies of transfer learning in a comprehensive way, which may help readers have a better understanding of the current research status and ideas.