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 Concept Videos

You might also read

Related Articles

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

Sort by
Same author

Molecular typing and characterization of a novel genotype of EV-B93 isolated from Tibet, China.

PloS one·2020
Same author

Fecal microbiota transplantation from patients with autoimmune encephalitis modulates Th17 response and relevant behaviors in mice.

Cell death discovery·2020
Same author

Comparison of Histogram-Based Gaussian Analysis With and Without Noise Correction for the Characterization of Indeterminate Adrenal Nodules.

AJR. American journal of roentgenology·2020
Same author

Photo-induced specific intracellular release EGFR inhibitor from enzyme/ROS-dual sensitive nano-platforms for molecular targeted-photodynamic combinational therapy of non-small cell lung cancer.

Journal of materials chemistry. B·2020
Same author

SYNTHESIS OF 1,3,4-OXADIAZOLES AS SELECTIVE T-TYPE CALCIUM CHANNEL INHIBITORS.

Heterocycles·2020
Same author

BMP-2 Signaling and Mechanotransduction Synergize to Drive Osteogenic Differentiation via YAP/TAZ.

Advanced science (Weinheim, Baden-Wurttemberg, Germany)·2020

Related Experiment Video

Updated: Sep 20, 2025

Movement Retraining using Real-time Feedback of Performance
08:16

Movement Retraining using Real-time Feedback of Performance

Published on: January 17, 2013

13.5K

Revising Representation and Target Deviations for Accurate Human Pose Estimation.

Zian Zhang, Yongqiang Zhang, Yancheng Bai

    IEEE Transactions on Neural Networks and Learning Systems
    |May 22, 2025
    PubMed
    Summary
    This summary is machine-generated.

    This study introduces DRPose, a novel framework to address representation and target deviations in human pose estimation (HPE). DRPose improves performance by adapting to scale variations and optimizing heatmap targets for better accuracy.

    More Related Videos

    Estimation of Contact Regions Between Hands and Objects During Human Multi-Digit Grasping
    09:41

    Estimation of Contact Regions Between Hands and Objects During Human Multi-Digit Grasping

    Published on: April 21, 2023

    1.7K
    Author Spotlight: Insights into the Analysis of Human Interaction with 3D Virtual Objects
    06:36

    Author Spotlight: Insights into the Analysis of Human Interaction with 3D Virtual Objects

    Published on: October 18, 2024

    1.1K

    Related Experiment Videos

    Last Updated: Sep 20, 2025

    Movement Retraining using Real-time Feedback of Performance
    08:16

    Movement Retraining using Real-time Feedback of Performance

    Published on: January 17, 2013

    13.5K
    Estimation of Contact Regions Between Hands and Objects During Human Multi-Digit Grasping
    09:41

    Estimation of Contact Regions Between Hands and Objects During Human Multi-Digit Grasping

    Published on: April 21, 2023

    1.7K
    Author Spotlight: Insights into the Analysis of Human Interaction with 3D Virtual Objects
    06:36

    Author Spotlight: Insights into the Analysis of Human Interaction with 3D Virtual Objects

    Published on: October 18, 2024

    1.1K

    Area of Science:

    • Computer Vision
    • Machine Learning
    • Artificial Intelligence

    Background:

    • Heatmap-based human pose estimation (HPE) methods with a top-down paradigm dominate performance due to normalized instance scales and robust supervision.
    • Existing frameworks suffer from representation deviation (unified input size for various scales) and target deviation (fixed prior distribution for prediction error), leading to performance bottlenecks.

    Purpose of the Study:

    • To propose a novel framework, DRPose, to revise the inherent representation and target deviations in top-down HPE methods.
    • To enhance the accuracy and robustness of human pose estimation by addressing scale variations and optimizing prediction targets.

    Main Methods:

    • A scale-aware domain bridging (SDB) block is introduced to transfer feature maps from multiple scale-dependent domains into a unified intermediate domain using dynamic parameters, mitigating representation deviation.
    • A differentiable coordinate decoder (DCD) is presented to adaptively adjust heatmap target distributions in an end-to-end manner, addressing target deviation.

    Main Results:

    • The proposed DRPose framework significantly improves the performance of most existing HPE models with negligible additional computational cost.
    • DRPose achieves 77.1% AP on the COCO test-dev dataset, outperforming prior works with similar model complexity.

    Conclusions:

    • DRPose effectively revises representation and target deviations in heatmap-based HPE, leading to substantial performance gains.
    • The framework offers a significant advancement in human pose estimation accuracy and efficiency, demonstrating its potential for real-world applications.