DenseCrack: A Segmentation Dataset of Dense Micro-Cracks in Strain-Hardening Cementitious Composites

Published: 23 October 2025| Version 1 | DOI: 10.17632/f9pwcwvkvj.1
Contributors:
Guo Pengwei, Yi Bao

Description

DenseCrack is a high-resolution image dataset designed for semantic segmentation of dense micro-cracks in strain-hardening cementitious composites (SHCCs). The dataset provides fine-grained, pixel-level annotations of micro-cracks, enabling precise evaluation and benchmarking of crack detection and segmentation models. Train data: 410 Test data: 100 Image size: 512 × 512

Files

Institutions

  • Stevens Institute of Technology

Categories

Image Segmentation

Licence