Generative AI-driven edge-cloud system for intelligent road infrastructure inspection
Bibliographic record
Abstract
The rapid advancement of edge computing and artificial intelligence (AI) has transformed infrastructure inspection by enabling real-time monitoring of roads, bridges, and pipelines. However, high bandwidth consumption, latency, and limited interpretability remain key challenges. This paper presents a novel hybrid edge-cloud framework for intelligent road infrastructure inspection, combining lightweight AI on edge devices with generative AI in the cloud. The Edge AI Module, built on MobileNetV3, performs real-time anomaly detection and generates concise reports with GPS-tagged severity information. Anomalous data is selectively transmitted to the cloud, where advanced models—EfficientNet-B4, MiDaS DPT-Large, and T5-XL—refine classification, estimate depth, compute road quality metrics, and generate structured, actionable reports. The system is evaluated on two diverse datasets: RDD2022, a multinational road damage dataset, and UAV-PDD2023, a high-resolution aerial imagery dataset. Results demonstrate the framework's real-time capability, achieving an edge inference time of 30 to 50 ms and reducing bandwidth usage by 50 to 70%. Cloud processing provides fine-grained analysis and high accuracy in natural language reporting. This dual-tier architecture balances low-latency anomaly detection and in-depth analysis, providing a scalable and interpretable solution for large-scale infrastructure monitoring in dynamic environments.
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) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.004 | 0.002 |
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".