Related Experiment Video
Updated: May 21, 2025
13:44
Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
Published on: August 30, 2013
42.6K
Segnet unveiled: Robust image segmentation via rigorous K-fold cross-validation analysis
Ignatious K Pious1, R Srinivasan1
1Department of Computer Science and Engineering, Vel Tech Rangarajan Dr Sagunthala R&D Institute of Science and Technology, Chennai, India.
Summary
The SegNet model demonstrates reliable image segmentation performance across diverse datasets, achieving high Dice Coefficient and Intersection over Union (IOU) scores. K-fold cross-validation confirms its generalization capabilities for various applications.
Area of Science:
- Computer Vision
- Machine Learning
Background:
- Image segmentation is a fundamental task in computer vision.
- Applications span autonomous driving and medical imaging.
Purpose of the Study:
- To evaluate the performance of a SegNet-based image segmentation model.
- To ensure reliable segmentation across varied datasets.
Main Methods:
- Utilized five-fold and K-fold cross-validation for model assessment.
- Measured Intersection over Union (IOU), Dice Coefficient, Precision, Recall, Accuracy, and loss metrics.
Main Results:
- Consistently high performance with Dice Coefficients from 88.32% to 89.8%.
- Intersection over Union (IOU) scores ranged from 94.53% to 95.05%.
- Precision, Recall, and Accuracy metrics frequently exceeded 90%, indicating model dependability.
Conclusions:
- K-fold cross-validation enhances validation reliability for the SegNet model.
- The model demonstrates strong generalization across datasets, suitable for practical image segmentation tasks.
