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

Related Concept Videos

Vision

Vision

Vision is the result of light being detected and transduced into neural signals by the retina of the eye. This information is then further analyzed and interpreted by the brain. First, light enters the front of the eye and is focused by the cornea and lens onto the retina—a thin sheet of neural tissue lining the back of the eye. Because of refraction through the convex lens of the eye, images are projected onto the retina upside-down and reversed.

You might also read

Related Articles

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

Sort by

Same author

Hybrid CNN-GNN architectures with distributed training for heathland plant classification.

Environmental monitoring and assessment·2026

Same author

Cell-matrix feedback controls stretch-induced cellular memory and fibroblast activation.

Proceedings of the National Academy of Sciences of the United States of America·2025

Same author

Expansion limits of meshed split-thickness skin grafts.

Acta biomaterialia·2024

Same author

CAC: Confidence-Aware Co-Training for Weakly Supervised Crack Segmentation.

Entropy (Basel, Switzerland)·2024

Same author

Integrating POLE/POLD1 mutated for immunotherapy treatment planning of advanced stage non-small cell lung cancer.

Thoracic cancer·2023

Same author

HAZMAT Vehicle Reidentification in Road Tunnels Based on the Fusion of Appearance and Spatiotemporal Information.

Computational intelligence and neuroscience·2023

Search research articles

Related Experiment Video

Updated: May 25, 2025

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

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

Published on: November 30, 2022

CrackCLIP: Adapting Vision-Language Models for Weakly Supervised Crack Segmentation.

Fengjiao Liang¹, Qingyong Li^1,2, Haomin Yu³

¹Key Laboratory of Big Data & Artificial Intelligence in Transportation (Beijing Jiaotong University), Ministry of Education, Beijing 100044, China.

Entropy (Basel, Switzerland)

|February 26, 2025

Summary

This study introduces CrackCLIP for efficient crack segmentation using minimal labels. The novel approach uses language prompts with a vision-language model to improve structural integrity assessments.

Keywords:

Contrastive Language–Image Pre-Training vision-language model weakly supervised crack segmentation

More Related Videos

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

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

Published on: April 14, 2023

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

Related Experiment Videos

Last Updated: May 25, 2025

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

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

Published on: November 30, 2022

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

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

Published on: April 14, 2023

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

Area of Science:

Computer Vision
Artificial Intelligence
Materials Science

Background:

Weakly supervised crack segmentation is vital for structural integrity assessment.
Pixel-level annotation is labor-intensive and impractical for real-world applications.
Existing methods struggle with labeling uncertainty.

Purpose of the Study:

To present CrackCLIP, a novel approach for weakly supervised crack segmentation.
To leverage language prompts and the Contrastive Language-Image Pre-Training (CLIP) model.
To enhance semantic context and improve segmentation accuracy with minimal annotation.

Main Methods:

Utilizing gradient-based class activation maps for coarse pseudo-label generation.
Fine-tuning CLIP's frozen image encoders with linear adapters for crack segmentation.
Employing textual prompts with CLIP's frozen text encoder to extract semantic features.
Comparing text prompt features with visual patch token features for final segmentation.

Main Results:

CrackCLIP outperforms existing weakly supervised crack segmentation methods on benchmark datasets (Crack500, CFD, DeepCrack).
The adapted vision-language model demonstrates strong potential for crack feature learning.
The framework shows enhanced performance and generalization capabilities.

Conclusions:

CrackCLIP effectively addresses labeling uncertainty in crack segmentation.
The integration of language prompts significantly boosts segmentation performance.
The proposed method offers a promising direction for automated structural health monitoring.