SCL-GCN: Stratified Contrastive Learning Graph Convolution Network for pavement crack detection from mobile LiDAR point clouds
Bibliographic record
Abstract
Accurate pavement crack detection is important for routine maintenance of pavements and reduction of possible traffic accidents. Most existing rule- or learning-based point-level approaches cannot achieve high detection accuracy and efficiency owing to the disorderly arrangement, scattered intensities, diverse crack structures, large data volumes, and complex annotation of mobile laser scanning (MLS) point clouds. To address these issues, we developed SCL-GCN, a Stratified Contrastive Learning Graph Convolution Network with a novel dual-branch architecture for MLS-point cloud-based pavement crack detection. First, a multi-scale graph representation construction module was designed based on a stratification strategy. This module creates strengthened spaces for the raw pavement point cloud and its downsampled subset, from which adjacency matrices and initial representations are generated. The stratification strategy samples neighbors densely in the raw point clouds and sparsely in the downsampled subset to form the neighborhood for each point, utilizing long-range contexts to increase the effective receptive field while lowing the extra computation. Next, a graph feature contrastive learning module is proposed to take advantage of stratified features. This module supervises the learning process of the two branches to avoid learning bias caused by an imbalanced data distribution, promoting convergence and improving performance. The experimental results show that the developed SCL-GCN model outperforms state-of-the-art methods. With a training/testing ratio of only 1:6 and an overall training time of less than 70 min, the average precision, recall, and F1-score of the SCL-GCN reached 75.7%, 75.1%, and 75.2%, respectively.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".