Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Weakly Supervised Adversarial Learning for 3D Human Pose Estimation from Point Clouds.

Zihao Zhang, Lei Hu, Xiaoming Deng

    IEEE Transactions on Visualization and Computer Graphics
    |February 20, 2020
    PubMed
    Summary
    This summary is machine-generated.

    Related Concept Videos

    Three-Dimensional Force System:Problem Solving01:30

    Three-Dimensional Force System:Problem Solving

    1.3K
    A three-dimensional force system refers to a scenario in which three forces act simultaneously in three different directions. This type of problem is commonly encountered in physics and engineering, where it is necessary to calculate the resultant force on the system, which can then be used to predict or analyze the behavior of the object or structure under consideration.
    To solve a three-dimensional force system, first resolve each force into its respective scalar components. Do this using...
    1.3K

    You might also read

    Related Articles

    Articles linked to this work by shared authors, journal, and citation graph.

    Sort by
    Same author

    The contributions of ammonia oxidizing bacteria and archaea to nitrification-dependent N<sub>2</sub>O emission in alkaline and neutral purple soils.

    Scientific reports·2022
    Same author

    MEK1-dependent MondoA phosphorylation regulates glucose uptake in response to ketone bodies in colorectal cancer cells.

    Cancer science·2022
    Same author

    Interface engineering with porous graphene as deposition regulator of stable Zn metal anode for long-life Zn-ion capacitor.

    Journal of colloid and interface science·2022
    Same author

    Fragment-Based Dynamic Combinatorial Chemistry for Identification of Selective α-Glucosidase Inhibitors.

    ACS medicinal chemistry letters·2022
    Same author

    Insulin-binding protein-5 down-regulates the balance of Th17/Treg.

    Frontiers in immunology·2022
    Same author

    Hsa_circ_0094606 promotes malignant progression of prostate cancer by inducing M2 polarization of macrophages through PRMT1-mediated arginine methylation of ILF3.

    Carcinogenesis·2022
    Same journal

    Two-phase Impulse Fluid on Particle Flow Map.

    IEEE transactions on visualization and computer graphics·2026
    Same journal

    FGO-SLAM++: Real-time Geometry-Aware Gaussian SLAM with Continuous Opacity Field.

    IEEE transactions on visualization and computer graphics·2026
    Same journal

    Blue Noise Dithering for Reservoir-based Spatio-temporal Importance Resampling.

    IEEE transactions on visualization and computer graphics·2026
    Same journal

    ROS-GS: Relightable Outdoor Scenes With Gaussian Splatting.

    IEEE transactions on visualization and computer graphics·2026
    Same journal

    MesoSplats: Texture Synthesis with Gaussian Splatting.

    IEEE transactions on visualization and computer graphics·2026
    Same journal

    GLLA: A Unified Force-Directed Graph Layout Framework Supporting Local Adjustments.

    IEEE transactions on visualization and computer graphics·2026
    See all related articles

    This study introduces a novel weakly supervised adversarial learning framework for 3D human pose estimation from point clouds. It effectively uses 2D joint annotations to improve accuracy in augmented reality/virtual reality applications.

    Area of Science:

    • Computer Vision
    • Machine Learning
    • Human-Computer Interaction

    Background:

    • 3D human pose estimation is crucial for AR/VR but relies on costly 3D joint annotations.
    • Existing methods struggle with labor-intensive and error-prone 3D data annotation due to occlusions.
    • Easier acquisition of 2D human joint annotations on depth images presents an opportunity for weaker supervision.

    Purpose of the Study:

    • To develop a weakly supervised adversarial learning framework for accurate 3D human pose estimation from point clouds.
    • To leverage both 2D and 3D joint annotations, overcoming limitations of solely 3D-annotated datasets.
    • To reduce computational cost and improve the domain gap between point cloud inputs and 3D joint outputs.

    Main Methods:

    • A weakly supervised adversarial learning framework combining 2D and 3D joint annotations.

    Related Experiment Videos

  • Utilizing 2D Convolutional Neural Networks (CNN) for 2D joint extraction from depth images.
  • Employing a point cloud network to refine 3D pose estimation and bridge the domain gap.

    • Main Results:

    • Achieved accurate 3D human pose estimation from point clouds using the proposed framework.
    • Demonstrated state-of-the-art performance on the ITOP and EVAL datasets.
    • Showcased efficient computation through effective sampling point selection.

    Conclusions:

    • Weakly supervised adversarial learning effectively addresses the challenge of 3D human pose estimation from point clouds.
    • The proposed method successfully integrates 2D and 3D annotations for improved accuracy and efficiency.
    • This approach offers a viable solution for AR/VR applications requiring precise 3D human pose recovery.