Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: May 22, 2026

Automated Joint Space Detection Improves Bone Segmentation Accuracy

Automated Joint Space Detection Improves Bone Segmentation Accuracy

Published on: November 28, 2025

Pose Guided Unsupervised Domain Adaptation for Human Body Part Segmentation.

Arindam Dutta, Rohit Lal, Yash Garg

IEEE Transactions on Image Processing : a Publication of the IEEE Signal Processing Society

|May 20, 2026

Summary

We developed POSTURE, a novel pose-guided method for human body part segmentation. It significantly improves accuracy on new datasets by using anatomical pose information for domain adaptation.

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

Smartphone-based offline AI for multi-disease retinal screening: Real-world accuracy.

European journal of ophthalmology·2026

Same author

Investigating Effect of Dimensional Variance on Separation of Glomerular Ultrafiltrate in a Microfluidic Environment.

IEEE transactions on nanobioscience·2026

Same author

Epoxyazadiradione, a neem-derived limonoid exhibits activities against pancreatic cancer through modulation of inflammatory molecules, lncRNAs, ROS, and EMT.

Biochimica et biophysica acta. General subjects·2026

Same author

SuryaBench: Benchmark Dataset for Advancing Machine Learning in Heliophysics and Space Weather Prediction.

Scientific data·2026

Same author

Sentinel Cerebral Embolic Protection in Patients with Transcatheter Aortic Valve Replacement: A Meta-analysis of Randomized Controlled Trials and Trial Sequential Analysis.

Cardiology in review·2025

Same author

Acute heart failure in patients with hematologic malignancies: A nationwide analysis 2016-2021.

SAGE open medicine·2025

Area of Science:

Computer Vision
Machine Learning
Medical Image Analysis

Background:

Current human body part segmentation algorithms struggle with domain shifts, leading to performance degradation.
End-to-end supervised methods are sensitive to variations in data distribution.

Purpose of the Study:

To introduce POSTURE (Pose Guided Unsupervised Domain Adaptation for Human Body Part Segmentation), a method to enhance segmentation performance on unlabeled target data.
To address the limitations of existing domain adaptive methods by incorporating anatomical priors.

Main Methods:

POSTURE utilizes a pseudo-labelling approach guided by anatomical pose keypoints.
It leverages the underlying structure of the human body to drive the domain adaptation process.
The method is extended to source-free settings (SF-POSTURE) for privacy and computational efficiency.

Related Experiment Videos

Last Updated: May 22, 2026

Automated Joint Space Detection Improves Bone Segmentation Accuracy

Automated Joint Space Detection Improves Bone Segmentation Accuracy

Published on: November 28, 2025

Main Results:

POSTURE achieved an average improvement of 8% over state-of-the-art domain adaptive semantic segmentation methods.
Performance gains were demonstrated across three benchmark datasets.
SF-POSTURE maintained performance with negligible degradation in source-free scenarios.

Conclusions:

POSTURE offers a robust solution for human body part segmentation under domain shifts.
Incorporating anatomical guidance provides a strong inductive prior for effective domain adaptation.
The source-free extension enhances the method's practicality and applicability.