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 Video

Updated: May 14, 2026

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments
03:31

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments

Published on: December 15, 2023

Semantic priors and virtual outlier synthesis enable parameter-efficient open-world object detection.

Jiaming Gu1, Yehui Zheng1, Yuzhou Liu1

  • 1Department of Computer Science, Guangdong University of Science and Technology, Dongguan, 523000, Guangdong, China.

Scientific Reports
|May 12, 2026
PubMed
Summary

Related Concept Videos

Difference from Background: Limit of Detection01:05

Difference from Background: Limit of Detection

The limit of detection (LOD) is the smallest amount of analyte that can be distinguished from the background noise. The LOD value corresponds to the concentration at which the analyte signal is three times larger than the standard deviation of the blank signal. Below this value, the analyte signal cannot be differentiated from the background noise. It is calculated by dividing the calibration slope by 3 times the standard deviation of the blank signals.
The LOD indicates the presence or absence...

You might also read

Related Articles

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

Sort by
Same author

Bufalin suppresses breast cancer bone metastasis through targeting SEC13-mediated osteoclastogenesis.

Phytomedicine : international journal of phytotherapy and phytopharmacology·2026
Same author

Longitudinal associations among insomnia, stigma, social support, and depressive symptoms in tuberculosis: evidence from RI-CLPM and LGCM.

BMC public health·2026
Same author

Association of the single-point insulin sensitivity estimator with arterial stiffness: a cross-sectional analysis.

Frontiers in endocrinology·2026
Same author

Designing Strong, Tough, Fire-Retardant and Self-Healing Elastomers with Phosphorus/Nitrogen- and Biphenyl-Containing Segments.

Advanced materials (Deerfield Beach, Fla.)·2026
Same author

The role of protein palmitoylation in disease pathogenesis and therapeutic innovation.

Annals of medicine·2026
Same author

Exploring the Efficacy of Immunotherapy in Combination with Surgical Resection in High-risk Muscle-invasive Bladder Cancer.

Iranian journal of allergy, asthma, and immunology·2026
This summary is machine-generated.

This study introduces Parameter-Efficient Open-World Object Detection (PE-OWOD), a novel framework that efficiently adapts models for discovering new objects without full retraining. PE-OWOD significantly improves performance on unknown objects while reducing computational costs.

Area of Science:

  • Computer Vision
  • Artificial Intelligence
  • Machine Learning

Background:

  • Open-world object detection necessitates identifying known and unknown object categories.
  • Full model fine-tuning in dynamic environments leads to overfitting on known classes and reduced sensitivity to novel objects, incurring high computational expenses.

Purpose of the Study:

  • To propose a Parameter-Efficient Open-World Object Detection (PE-OWOD) framework for efficient retraining and adaptation in dynamic object detection scenarios.
  • To enhance the model's ability to discover and detect previously unseen objects while maintaining computational efficiency.

Main Methods:

  • Developed the PE-OWOD framework, which freezes the backbone and encoder to preserve visual priors.
  • Introduced compact Residual adapters injected solely into the decoder for task adaptation.
Keywords:
Computer visionIncremental learningOpen-world object detectionParameter-efficient fine-tuningTransformerVirtual outlier synthesis

Related Experiment Videos

Last Updated: May 14, 2026

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments
03:31

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments

Published on: December 15, 2023

  • Incorporated Visual Open Space (VOS) to establish a clear decision boundary for unknown objects, with optional semantic initialization.

    • Main Results:

    • Achieved 64.7% Unknown Recall on MS-COCO benchmarks, outperforming fully tuned baselines.
    • Reduced GPU memory usage by 86% by updating less than 27% of the model parameters.
    • Demonstrated significant efficiency advantages and robust performance in open-world detection.

    Conclusions:

    • Parameter-efficient adaptation is a viable and effective strategy for open-world object detection.
    • The PE-OWOD framework offers a reliable and computationally efficient solution for detecting both known and unknown objects.
    • This approach overcomes the limitations of full fine-tuning in dynamic and open-ended visual recognition tasks.