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

The alkaloid-rich extract of Piper sintenense Hatusima alleviates gouty inflammatory response via Nrf2-mediated antioxidant induction.

Inflammopharmacology·2026
Same author

Psychiatric events induced by roflumilast: a real-world pharmacovigilance study of the FDA Adverse Event Reporting System database.

Frontiers in psychiatry·2026
Same author

A real-world pharmacovigilance study of romidepsin based on FDA adverse event reporting system database.

Frontiers in oncology·2026
Same author

Gaussian Process Regression for Tail Vehicle Departure Time Prediction at Signalized Intersections Using UAV Trajectory Data.

Sensors (Basel, Switzerland)·2026
Same author

‌A century-scale dataset of crop yield and agronomic trait evolution under field management in China.

Scientific data·2026
Same author

Emotion regulation interventions for autistic children and adolescents: A panoramic comparison through systematic review and network meta-analysis.

Clinical psychology review·2026

Related Experiment Video

Updated: Oct 10, 2025

Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment
08:25

Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment

Published on: May 7, 2019

9.2K

Learning From Pixel-Level Label Noise: A New Perspective for Semi-Supervised Semantic Segmentation.

Rumeng Yi, Yaping Huang, Qingji Guan

    IEEE Transactions on Image Processing : a Publication of the IEEE Signal Processing Society
    |December 15, 2021
    PubMed
    Summary

    This study introduces a novel graph-based framework for semi-supervised semantic segmentation, effectively handling pixel-level label noise from weak supervisions. The method achieves state-of-the-art results, outperforming fully-supervised models in some cases.

    More Related Videos

    Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
    04:48

    Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

    Published on: November 30, 2022

    3.0K
    Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application
    05:56

    Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application

    Published on: April 14, 2023

    2.7K

    Related Experiment Videos

    Last Updated: Oct 10, 2025

    Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment
    08:25

    Combining Eye-tracking Data with an Analysis of Video Content from Free-viewing a Video of a Walk in an Urban Park Environment

    Published on: May 7, 2019

    9.2K
    Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
    04:48

    Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

    Published on: November 30, 2022

    3.0K
    Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application
    05:56

    Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application

    Published on: April 14, 2023

    2.7K

    Area of Science:

    • Computer Vision
    • Machine Learning
    • Artificial Intelligence

    Background:

    • Semi-supervised semantic segmentation leverages limited pixel-level annotations (strong supervision) and abundant image-level annotations (weak supervision).
    • Existing methods often generate noisy pixel-level labels from weak supervision, hindering performance.
    • Current noisy label learning techniques are primarily designed for image-level tasks, failing to capture spatial relationships within images.

    Purpose of the Study:

    • To address the challenge of pixel-level label noise in semi-supervised semantic segmentation.
    • To propose a novel framework for detecting and correcting noisy pixel-level labels derived from weak supervision.
    • To advance the state-of-the-art in semi-supervised semantic segmentation by effectively utilizing both strong and weak supervision.

    Main Methods:

    • Formulating the task as learning with pixel-level label noise.
    • Developing a graph-based framework for label noise detection and correction.
    • Utilizing Class Activation Maps (CAM) for initial label generation, a clean model for noise detection via cross-entropy loss, and a superpixel-based graph with a Graph Attention Network (GAT) for label correction.

    Main Results:

    • The proposed method achieves state-of-the-art performance on benchmark datasets including PASCAL VOC 2012, PASCAL-Context, MS-COCO, and Cityscapes.
    • Experimental results demonstrate the effectiveness of the graph-based approach in handling pixel-level label noise.
    • In certain scenarios, the semi-supervised method surpasses the performance of fully-supervised models on PASCAL VOC 2012 and MS-COCO.

    Conclusions:

    • The proposed graph-based label noise detection and correction framework is highly effective for semi-supervised semantic segmentation.
    • This approach successfully mitigates the impact of noisy labels generated from weak supervision.
    • The method offers a promising direction for improving semantic segmentation accuracy with limited annotated data.