Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Benefiting from Quantum? A Comparative Study of Q-Seg, Quantum-Inspired Techniques, and U-Net for Crack Segmentation

Akshaya Srinivasan, Alexander Geng, Antonio Macaluso, Maximilian Kiefer-Emmanouilidis, Ali Moghiseh

TL;DR

This study evaluates the performance of quantum and quantum-inspired methods compared to classical models for crack segmentation using annotated gray-scale image patches of concrete samples to indicate that quantum-inspired and quantum methods offer a promising alternative for image segmentation, particularly for complex crack patterns, and could be applied in near-future applications.

Abstract

Exploring the potential of quantum hardware for enhancing classical and real-world applications is an ongoing challenge. This study evaluates the performance of quantum and quantum-inspired methods compared to classical models for crack segmentation. Using annotated gray-scale image patches of concrete samples, we benchmark a classical mean Gaussian mixture technique, a quantum-inspired fermion-based method, Q-Seg a quantum annealing-based method, and a U-Net deep learning architecture. Our results indicate that quantum-inspired and quantum methods offer a promising alternative for image segmentation, particularly for complex crack patterns, and could be applied in near-future applications.

Benefiting from Quantum? A Comparative Study of Q-Seg, Quantum-Inspired Techniques, and U-Net for Crack Segmentation

TL;DR

This study evaluates the performance of quantum and quantum-inspired methods compared to classical models for crack segmentation using annotated gray-scale image patches of concrete samples to indicate that quantum-inspired and quantum methods offer a promising alternative for image segmentation, particularly for complex crack patterns, and could be applied in near-future applications.

Abstract

Exploring the potential of quantum hardware for enhancing classical and real-world applications is an ongoing challenge. This study evaluates the performance of quantum and quantum-inspired methods compared to classical models for crack segmentation. Using annotated gray-scale image patches of concrete samples, we benchmark a classical mean Gaussian mixture technique, a quantum-inspired fermion-based method, Q-Seg a quantum annealing-based method, and a U-Net deep learning architecture. Our results indicate that quantum-inspired and quantum methods offer a promising alternative for image segmentation, particularly for complex crack patterns, and could be applied in near-future applications.

Paper Structure

This paper contains 11 sections, 4 equations, 4 figures, 1 table.

Table of Contents

  1. Introduction
  2. Segmentation Techniques
  3. Mean Gaussian Mixture
  4. Quantum-Inspired Hamiltonian
  5. Q-Seg: Unsupervised Quantum Algorithm
  6. U-Net
  7. Dataset and Metrics for Segmentation Analysis
  8. Evaluation Metrics
  9. Boundary Proximity Metric
  10. Results and Discussion
  11. Acknowledgements

Figures (4)

  • Figure 1: Overview of crack segmentation motivation and methodology: (a) Cracks on roads illustrating real-world infrastructure challenges, (b) Results from the QI approach, accurately identifying crack locations using localized states tied to negative eigenvalues, and (c) Comparative pipeline of crack segmentation methods.
  • Figure 2: Sample images of cracks with corresponding masks.
  • Figure 3: Crack segmentation results from four different techniques: MGM, the QI Hamiltonian method, U-Net, and Q-Seg.
  • Figure 4: Visual comparison of segmentation results, including the standard confusion matrix (a), the confusion matrix post-BPM application (b), and an overlay of predicted crack masks against actual cracks before (c) and after BPM (d).