Image captioning for automated bridge inspection: a feasibility study

As the number of bridges grows in China, bridge inspection is very necessary to ensure public transport safety. With the development of various technologies in recent years, such as UAV (Unmanned Aerial Vehicle), computer vision, advanced sensing, artificial intelligence and so on, intelligent technologies in bridge inspection have developed rapidly and are gradually replacing traditional methods. Here we proposed parallel systems for bridge inspection, which introduces parallel theory into the field of bridge inspection to solve the problems of dataset shortage and special scene prediction. Based on the classification dataset (CCD) and parallel classification dataset (PCD), ConvNeXt and other networks are trained and compared. The final crack identification accuracy reached 99.22%. We believe that the framework proposed in this paper can improve the efficiency and accuracy of bridge inspection significantly.

Traditional bridge inspection methods, such as visual inspections and basic non-destructive tests, remain foundational but face limitations due to labor intensity and dependency on inspector expertise, often resulting in inconsistencies. Machine learning (ML) tools offer transformative solutions by enabling the analysis of large datasets to enhance damage detection accuracy, optimize maintenance schedules, and improve resource allocation. The objective of this study was to explore the integration of advanced inspection tools, such as Unmanned Aerial Vehicles (UAVs), digital imaging, and fiber optic sensors, with ML models to enhance data accuracy, decision-making, and inspection rating efficiency. A state-of-the-art review is conducted in this study on the application of ML techniques in bridge inspection and maintenance, encompassing 60 articles and reports published in the last decade. The results of this study show that ML’s ability to integrate advanced inspection tools to improve data accuracy, decision-making, and inspection efficiency. However, the existing challenges persist in data quality, model generalization, and the need for standardization approaches across diverse bridge conditions. The review provides insights into current methodologies, benefits, limitations, and future directions, emphasizing ML’s pivotal role in modernizing bridge health monitoring for a safer and more sustainable infrastructure network. The findings of this study are expected to assist transportation agencies in planning and operating predictive maintenance strategies, focusing on damage detection, maintenance optimization, and resource allocation.

PurposeThis paper covers the development of a novel defect model for concrete highway bridges. The proposed defect model is intended to facilitate the identification of bridge’s condition information (i.e. defects), improve the efficiency and accuracy of bridge inspections by supporting practitioners and even machines with digitalised expert knowledge, and ultimately automate the process.Design/methodology/approachThe research design consists of three major phases so as to (1) categorise common defect with regard to physical entities (i.e. bridge element), (2) establish internal relationships among those defects and (3) relate defects to their properties and potential causes. A mixed-method research approach, which includes a comprehensive literature review, focus groups and case studies, was employed to develop and validate the proposed defect model.FindingsThe data collected through the literature and focus groups were analysed and knowledge were extracted to form the novel defect model. The defect model was then validated and further calibrated through case study. Inspection reports of nearly 300 bridges in China were collected and analysed. The study uncovered the relationships between defects and a variety of inspection-related elements and represented in the form of an accessible, digitalised and user-friendly knowledge model.Originality/valueThe contribution of this paper is the development of a defect model that can assist inexperienced practitioners and even machines in the near future to conduct inspection tasks. For one, the proposed defect model can standardise the data collection process of bridge inspection, including the identification of defects and documentation of their vital properties, paving the path for the automation in subsequent stages (e.g. condition evaluation). For another, by retrieving rich experience and expert knowledge which have long been reserved and inherited in the industrial sector, the inspection efficiency and accuracy can be considerably improved.

Unmanned aerial vehicles (UAVs) are increasingly leveraged to perform infrastructure inspection tasks, especially with the support of rapidly evolving AI algorithms and hardware in recent years. While the integration of UAV and AI techniques enhances the efficiency and effectiveness of infrastructure inspection, it also raises security concerns due to the potential vulnerabilities existing in the underlying AI models. In this paper, we propose to investigate and discover these vulnerabilities with the case study on bridge inspection. In particular, we designed a two-stage approach that can construct effective adversarial perturbations that make the UAV miss the detection of risk-prone regions during the inspection. Spatial constraints, physical limits, as well as dynamic environmental changes are taken into consideration in our approach to make it practical in the physical world. We evaluate our approach using the COCO-Bridge dataset. Our experimental results demonstrate the effectiveness of our approach in both white-box attack and black-box attack settings.

Purpose With the deep integration of information technologies into urban governance, smart communities have emerged as pivotal platforms for advancing sustainable urban development. However, existing research has not offered a systematic analysis or clear presentation of the field’s academic evolution and thematic structure. This study examines the literature on smart communities published between 2000 and 2024. Employing data analysis and visualization tools, it aims to trace the evolution and development trends of smart community research, map its core themes and their interrelationships, and provide actionable insights for policymaking and practical implementation. Design/methodology/approach Based on 2,347 publications indexed in the Web of Science from 2000 to 2024, this study employed CiteSpace and VOSviewer to conduct co-citation analysis, keyword co-occurrence mapping, national collaboration network analysis, author and institutional contribution assessment, burst detection, and hotspot term analysis. The literature screening adhered to predefined publication-type criteria and citation-count thresholds to ensure that the results were representative and reliable. Findings This study, through literature analysis and data visualization in the field of smart communities, yields the following principal conclusions. First, the application of digital twin technology in optimizing smart community resources has attracted growing attention, demonstrating considerable potential in urban management, infrastructure maintenance, and resident services. Second, as technology advances, digitaltwin applications are evolving towards greater precision and efficiency, particularly by deepening their support for resource allocation and decision-making processes. Finally, the future development of smart communities will increasingly depend on the deep integration of digital twins with other cutting-edge technologies, thereby driving intelligent management and optimization of community resources. Research limitations This literature repository excludes grey literature and non-English publications, potentially underestimating the representativeness of grassroots innovation. Furthermore, the temporal analysis was constrained by the citation-lag effects of publications from 2000 to 2024. Practical implications This study proposes a decision-support toolkit tailored for municipal planners and policymakers. The toolkit comprises three core intervention strategies: multiscale environmental sensing, participatory governance protocols, and regenerative technology pathways. These measures are designed to advance the implementation of the United Nations Sustainable Development Goal 11: Sustainable Cities and Communities. Originality/value By explicitly defining and applying the “thematic knowledge framework,” this paper offers a concise roadmap for the evolution of smart community research. It also provides precise guidance for designing and implementing community development strategies that align with Sustainable Development Goals.

Rising traffic volumes and natural disasters such as earthquakes and typhoons accelerate bridge deterioration and intensify public concern over aging infrastructure. With aging bridges continuing to operate at peak capacity, there is an urgent need for an efficient, consistent, and reliable method of bridge-damage inspection. The DER and U (Degree of degradation, Extension, Relevance, and Urgency) visual inspection method, currently widely used in bridge inspections, is significantly influenced by high inspector turnover rates and inspector subjectivity, leading to inconsistencies in inspection results and recommendations. To address this need, an augmented reality (AR)-based mobile application was developed in this study to enhance bridge inspection training and improve data management. Using Google ARCore’s plane identification technology, the developed application accurately overlays virtual deterioration models onto real-world surfaces, providing inspectors with data that may be applied directly to objective decision making. The developed AR tool not only boosts trainee engagement and understanding but also includes a robust image collection and storage feature that enables inspectors to systematically capture, classify, and archive bridge condition images. This data repository supports the consistent, long-term tracking of structural deterioration and facilitates degradation-pattern analysis, ultimately contributing to more-effective bridge infrastructure maintenance planning and lifecycle management. The results of this study suggest the developed AR-based application has strong potential to modernize bridge inspection processes, offering the dual benefits of enhanced training and reliable data collection.

Image captioning refers to the process of generating a textual description that describes objects and activities present in a given image. It connects two fields of artificial intelligence, computer vision, and natural language processing. Computer vision and natural language processing deal with image understanding and language modeling, respectively. In the existing literature, most of the works have been carried out for image captioning in the English language. This article presents a novel method for image captioning in the Hindi language using encoder–decoder based deep learning architecture with efficient channel attention. The key contribution of this work is the deployment of an efficient channel attention mechanism with bahdanau attention and a gated recurrent unit for developing an image captioning model in the Hindi language. Color images usually consist of three channels, namely red, green, and blue. The channel attention mechanism focuses on an image’s important channel while performing the convolution, which is basically to assign higher importance to specific channels over others. The channel attention mechanism has been shown to have great potential for improving the efficiency of deep convolution neural networks (CNNs). The proposed encoder–decoder architecture utilizes the recently introduced ECA-NET CNN to integrate the channel attention mechanism. Hindi is the fourth most spoken language globally, widely spoken in India and South Asia; it is India’s official language. By translating the well-known MSCOCO dataset from English to Hindi, a dataset for image captioning in Hindi is manually created. The efficiency of the proposed method is compared with other baselines in terms of Bilingual Evaluation Understudy (BLEU) scores, and the results obtained illustrate that the method proposed outperforms other baselines. The proposed method has attained improvements of 0.59%, 2.51%, 4.38%, and 3.30% in terms of BLEU-1, BLEU-2, BLEU-3, and BLEU-4 scores, respectively, with respect to the state-of-the-art. Qualities of the generated captions are further assessed manually in terms of adequacy and fluency to illustrate the proposed method’s efficacy.

A big size of bridge data is being constantly generated—from a variety of distributed sources and in heterogeneous formats—in the form of national bridge inventory (NBI) data, national bridge elements (NBE) data, bridge inspection and maintenance reports, traffic data, etc. Big bridge data analytics open unprecedented opportunities to leverage this large amount of data for improved bridge deterioration prediction and understanding, and enhanced maintenance decision making. In this paper, the authors propose a big bridge data analytics framework that consists of: (1) semantic data/information extraction and integration, (2) semantic data/information analysis and mining, and (3) semantic data/information retrieval and visualization. The big size of the data, in addition to it being scattered and heterogeneity of meaning and format, adds to the challenge of data integration and analysis. As such, at the cornerstone of this framework is an ontology of bridge deterioration knowledge, which aims to facilitate the semantic integration and analysis of the data based on content and domain-specific meaning, and to bridge the terminology gap across different sources. This paper focuses on providing an overview of the framework for big bridge data analytics and presenting the proposed bridge deterioration ontology as the first step towards implementing this framework.

This research examines the management of school infrastructure maintenance in improving learning quality at SDIT Subulussalam Raman Utara, East Lampung. The study aims to analyze the infrastructure maintenance management process, assess learning quality improvement, and identify implementation barriers and solutions. The research employs a qualitative approach with descriptive analytical design, involving nine informants selected through purposive sampling. Data collection was conducted through participatory observation, in-depth interviews, and documentation studies. The findings reveal the implementation of maintenance management through three stages: collaborative-based planning, implementation with routine and periodic maintenance patterns, and systematic evaluation. Infrastructure maintenance positively correlates with learning quality in psychological, pedagogical, and administrative dimensions. Major constraints include budget limitations, structural damage, lack of technological facilities, and low digital competencies among educators. Mitigation strategies encompass developing external collaborative networks, enhancing educator capacity, reorienting budget priorities, and implementing participatory maintenance approaches.

The aim of this integrative review is to critically appraise and synthesise empirical evidence on the clinical applications, outcomes, and implications of generative artificial intelligence in nursing practice. Integrative review following Whittemore and Knafl's five-stage framework. Systematic searches were performed for peer-reviewed articles and book chapters published between 1 January 2018 and 30 June 2025. Two reviewers independently screened titles/abstracts and full texts against predefined inclusion/exclusion criteria focused on generative artificial intelligence tools embedded in nursing clinical workflow (excluding nursing education-only applications). Data were extracted into a standardised matrix and appraised for quality using design-appropriate checklists. Guided by Whittemore and Knafl's integrative review framework, a constant comparative analysis was applied to derive the main themes and subthemes. CINAHL, MEDLINE, and Embase. Included literature was a representative mix of single-group quality improvement pilots, mixed-method usability and feasibility studies, randomised controlled trials, qualitative descriptive and phenomenological studies, as well as preliminary and proof-of-concept observational research. Four overarching themes emerged: (1) Workflow Integration and Efficiency, (2) AI-Augmented Clinical Reasoning, (3) Patient-Facing Communication and Education, and (4) Role Boundaries, Ethics and Trust. Generative artificial intelligence holds promise for enhancing nursing efficiency, supporting clinical decision making, and extending patient communication. However, consistent human validation, ethical boundary setting, and more rigorous, longitudinal outcome and equity evaluations are essential before widespread clinical adoption. Although generative artificial intelligence could reduce nurses' documentation workload and routine decision-making burden, these gains cannot be assumed. Safe and effective integration will require rigorous nurse training, robust governance, transparent labelling of AI-generated content, and ongoing evaluation of both clinical outcomes and equity impacts. Without these safeguards, generative artificial intelligence risks introducing new errors and undermining patient safety and trust. PRISMA 2020.

Host-microbiome interactions are crucial in maintaining physiological homeostasis and influencing disease progression. While traditional microbiome research, focused on microbial diversity and abundance, provided essential frameworks, the integration of multi-omics approaches established a far more comprehensive, systems-level understanding of microbiome functionality. However, significant methodological challenges in multi-omics data integration remain, including inconsistent sample coverage, heterogeneous data formats, and complex downstream analytical workflows. These challenges impact the reproducibility and reliability of results and underscore the critical requirement for developing standardized, systematic workflows. Thus, we developed the EasyMultiProfiler (EMP) workflow, a streamlined and efficient analytical framework for multi-omics data analysis. EMP provides a comprehensive infrastructure based on the SummarizedExperiment and MultiAssayExperiment classes, creating a unified framework for storing and analyzing omics data. The framework's architecture comprises five interconnected functional modules: data extraction, preparation, support, analysis, and visualization. These modules are smoothly integrated into natural language-style analytical workflows, offering users an efficient and standardized solution for multi-omics analysis. EMP addresses critical challenges in multi-omics data analysis, including data integration, workflow standardization, and result reproducibility. The platform's modular architecture and intuitive interface provide researchers and clinicians with a robust, flexible workflow to systematically extract biologically relevant insights from complex multi-omics datasets.

In this research, we aim to propose an image sharpening method to specify concrete cracks from degraded images captured by a camera on a moving object. We have tried new method "convoluted PSF" that is a convolution of PSF (out-of-focus blur) and PSF (motion blur). We worked on blurred image sharpening by means of deconvolution using convoluted PSF instead of PSF (out-of-focus blur) and PSF (motion blur) step by step. And bridge inspection sites could make this method more efficient with some speckled texture on walls for estimation of PSF (motion blur). The effectiveness of this proposed method is confirmed by deconvolution of degraded images acquired by a digital camera used in actual bridge inspections. This study is expected to help realize social infrastructure maintenance using a wide range of robotic technologies, and to solve the future labor shortage and shortage of engineers.

Nasogastric tubes (NGTs) are feeding tubes that are inserted through the nose into the stomach to deliver nutrition or medication. If not placed correctly, they can cause serious harm, even death to patients. Recent AI developments demonstrate the feasibility of robustly detecting NGT placement from Chest X-ray images to reduce risks of sub-optimally or critically placed NGTs being missed or delayed in their detection, but gaps remain in clinical practice integration. In this study, we present a human-centered approach to the problem and describe insights derived following contextual inquiry and in-depth interviews with 15 clinical stakeholders. The interviews helped understand challenges in existing workflows, and how best to align technical capabilities with user needs and expectations. We discovered the trade-offs and complexities that need consideration when choosing suitable workflow stages, target users, and design configurations for different AI proposals. We explored how to balance AI benefits and risks for healthcare staff and patients within broader organizational, technical, and medical-legal constraints. We also identified data issues related to edge cases and data biases that affect model training and evaluation; how data documentation practices influence data preparation and labelling; and how to measure relevant AI outcomes reliably in future evaluations. We discuss how our work informs design and development of AI applications that are clinically useful, ethical, and acceptable in real-world healthcare services.

Persistence images, derived from topological data analysis, emerge as a powerful tool for visualizing and mitigating biases in radiological data interpretation and AI model development. This technique transforms complex topological features into stable, interpretable representations, offering unique insights into medical imaging data structure. By providing intuitive visualizations, persistence images enable the identification of subtle structural differences and potential biases in data acquisition, interpretation, and AI model training. Persistence images can also facilitate stratified sampling, matching statistics, and noise filtration, enhancing the accuracy and equity of radiological analysis. Despite challenges in computational complexity and workflow integration, persistence images show promise in developing more accurate, equitable, and trustworthy AI systems in radiology, potentially improving patient outcomes and personalized healthcare delivery.

There are over 590,000 bridges dispersed across the roadway network that stretches across the United States alone. Each bridge with a length of 20 feet or greater must be inspected at least once every 24 months, according to the Federal Highway Act (FHWA) of 1968. This research developed an artificial intelligence (AI)-based framework for bridge and road inspection using drones with multiple sensors collecting capabilities. It is not sufficient to conduct inspections of bridges and roads using cameras alone, so the research team utilized an infrared (IR) camera along with a high-resolution optical camera. In many instances, the IR camera can provide more details to the interior structural damages of a bridge or a road surface than an optical camera, which is more suitable for inspecting damages on the surface of a bridge or a road. In addition, the drone inspection system is equipped with a minicomputer that runs Machine Learning algorithms. These algorithms enable autonomous drone navigation, image capture of the bridge or road structure, and analysis of the images. Whenever any damage is detected, the location coordinates are saved. Thus, the drone can self-operate and carry out the inspection process using advanced AI algorithms developed by the research team. The experimental results reveal the system can detect potholes with an average accuracy of 84.62% using the visible light camera and 95.12% using a thermal camera. This developed bridge and road inspection framework can save time, money, and lives by automating and having drones conduct major inspection operations in place of humans.