Integrating Transformer Segmentation and Morphological Analysis for Multi-Crack Characterization and Visual Equivalent Strain Assessment in Rubberized ECC
- 1Guizhou Power Grid Co., Ltd., Guiyang 550029, China
- 2School of Civil and Transportation Engineering, Guangdong University of Technology, Guangzhou 510006, China
Abstract
Deep-learning semantic segmentation and morphological analysis offer a non-contact route to characterizing microcracks in cementitious composites, yet linking pixel-level masks to physical geometry and tensile response remains challenging. This study establishes an integrated lightweight segmentation and geometric-quantification workflow for rubberized engineered cementitious composites (ECCs) incorporating 0–50% powdered rubber. We propose EfficientSegLite, a supervised binary segmentation architecture combining a hierarchical SegFormer-B0 encoder, a channel-compressed U-Net-style decoder, lightweight atrous spatial pyramid pooling, and depthwise separable convolutions, trained via a two-stage Dice–OHEM–Edge strategy. On 1440 ECC tensile images, it achieved 93.40% mIoU and 93.24% F1 score with only 3.96 M parameters and 734.9 M FLOPs. Deterministic morphological and distance-transform algorithms then convert the masks into calibrated physical descriptors, achieving a mean relative crack-width error of 3.65% against optical microscopy across eight benchmark cracks. A crack-area-based visual equivalent strain (ε_v) correlated strongly with extensometer strain (Pearson’s r > 0.97, R2 > 0.94); however, regression slopes of 0.368–1.046 and uncalibrated mean absolute errors of 0.03–2.04 percentage points indicate that quantitative strain estimation requires mixture-specific calibration. A tool-calling large language model agent parses user instructions and orchestrates the deterministic computation tools, generating traceable inspection reports.
Related
- Rheological and Morphological Characterization of Asphalt Cement Hybrid-Modified with Waste-Derived Cellulose Fibers and Crumb Rubber
- Comparative Analysis of Starch-Based Films Functionalized with Bixa orellana Fruit and Leaf Extracts: A Multiscale Structural, Morphological, and Optical Characterization
- Morphological, genetic, and phenolic characterization of Baccharis glutinosa and Baccharis salicifolia and their application in the biogenic synthesis of silver nanoparticles
- Mercury(II) Coordination Compounds Based on Pyridine-2,5-Dicarboxylic Acid: Synthesis, Crystal Structure, Morphological Characterization, and Biological Activities.
- Oligothiophene-g-oligo(D,L-lactide) “Rod-Graft-Coil” Co-Oligomer: Self-Assembly Modulation by Processing and Formed Films Surface Morphological Diversity Characterization †
BibTeX
@article{jia2026integrating,
title = {{Integrating Transformer Segmentation and Morphological Analysis for Multi-Crack Characterization and Visual Equivalent Strain Assessment in Rubberized ECC}},
author = {Junqing Jia and Tengxiang Su and Jinghao Huang and Ge Zhang and Jiaxiang Lin},
journal = {Polymers},
year = {2026},
month = oct,
doi = {10.3390/polym18212657},
url = {https://www.mdpi.com/2073-4360/18/21/2657},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Integrating Transformer Segmentation and Morphological Analysis for Multi-Crack Characterization and Visual Equivalent Strain Assessment in Rubberized ECC
AU - Junqing Jia
AU - Tengxiang Su
AU - Jinghao Huang
AU - Ge Zhang
AU - Jiaxiang Lin
PY - 2026
DA - 2026/10/26
JO - Polymers
DO - 10.3390/polym18212657
UR - https://www.mdpi.com/2073-4360/18/21/2657
AB - Deep-learning semantic segmentation and morphological analysis offer a non-contact route to characterizing microcracks in cementitious composites, yet linking pixel-level masks to physical geometry and tensile response remains challenging. This study establishes an integrated lightweight segmentation and geometric-quantification workflow for rubberized engineered cementitious composites (ECCs) incorporating 0–50% powdered rubber. We propose EfficientSegLite, a supervised binary segmentation architecture combining a hierarchical SegFormer-B0 encoder, a channel-compressed U-Net-style decoder, lightweight atrous spatial pyramid pooling, and depthwise separable convolutions, trained via a two-stage Dice–OHEM–Edge strategy. On 1440 ECC tensile images, it achieved 93.40% mIoU and 93.24% F1 score with only 3.96 M parameters and 734.9 M FLOPs. Deterministic morphological and distance-transform algorithms then convert the masks into calibrated physical descriptors, achieving a mean relative crack-width error of 3.65% against optical microscopy across eight benchmark cracks. A crack-area-based visual equivalent strain (ε_v) correlated strongly with extensometer strain (Pearson’s r > 0.97, R2 > 0.94); however, regression slopes of 0.368–1.046 and uncalibrated mean absolute errors of 0.03–2.04 percentage points indicate that quantitative strain estimation requires mixture-specific calibration. A tool-calling large language model agent parses user instructions and orchestrates the deterministic computation tools, generating traceable inspection reports.
SN - 2073-4360
ER -