Accurate crime prediction is crucial for the proactive allocation of law enforcement resources and ensuring urban safety. A major challenge in achieving accurate predictions lies in identifying generalized patterns of criminal behavior from spatiotemporal features in crime data. Additionally, the inherent randomness and volatility of crime data at the spatiotemporal level introduce noise, which can mislead prediction models. While many effective spatiotemporal crime prediction methods have been proposed, most overlook this issue, reducing their ability to generalize. In this paper, we introduce a novel deep learning-based model, adaptive-GCNLSTM (Ada-GCNLSTM). Specifically, in the spatial feature extraction module, we enhance the model's ability to capture crime spatial distributions by leveraging graph convolutional networks to model spatial dependencies in conjunction with the maximum mean discrepancy to extract the universal features of crime data. We then incorporate a memory network based on long short-term memory network to capture the underlying relationships between temporal features. Through extensive experiments, our model demonstrates an average improvement of 11.7% in mean absolute error and 2.7% in root mean squared error across the three datasets, outperforming the best baseline model. These results underscore the effectiveness of our approach in enhancing crime prediction accuracy.
{"title":"Ada-GCNLSTM: An adaptive urban crime spatiotemporal prediction model","authors":"Miaoxuan Shan , Chunlin Ye , Peng Chen , Shufan Peng","doi":"10.1016/j.jnlssr.2024.11.003","DOIUrl":"10.1016/j.jnlssr.2024.11.003","url":null,"abstract":"<div><div>Accurate crime prediction is crucial for the proactive allocation of law enforcement resources and ensuring urban safety. A major challenge in achieving accurate predictions lies in identifying generalized patterns of criminal behavior from spatiotemporal features in crime data. Additionally, the inherent randomness and volatility of crime data at the spatiotemporal level introduce noise, which can mislead prediction models. While many effective spatiotemporal crime prediction methods have been proposed, most overlook this issue, reducing their ability to generalize. In this paper, we introduce a novel deep learning-based model, adaptive-GCNLSTM (Ada-GCNLSTM). Specifically, in the spatial feature extraction module, we enhance the model's ability to capture crime spatial distributions by leveraging graph convolutional networks to model spatial dependencies in conjunction with the maximum mean discrepancy to extract the universal features of crime data. We then incorporate a memory network based on long short-term memory network to capture the underlying relationships between temporal features. Through extensive experiments, our model demonstrates an average improvement of 11.7% in mean absolute error and 2.7% in root mean squared error across the three datasets, outperforming the best baseline model. These results underscore the effectiveness of our approach in enhancing crime prediction accuracy.</div></div>","PeriodicalId":62710,"journal":{"name":"安全科学与韧性(英文)","volume":"6 2","pages":"Pages 226-236"},"PeriodicalIF":3.7,"publicationDate":"2025-02-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143852015","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-14DOI: 10.1016/j.jnlssr.2024.11.004
David Javier Castro Rodriguez, Antonello A. Barresi, Micaela Demichela
Directive 2022/2557 from the European Commission aims to enhance the resilience of critical entities in Europe by integrating with existing European legislation, but it lacks explicit guidance on addressing vulnerabilities. Specifically, major hazard industries (MHIs) are critical infrastructures that face unique risks arising from the interactions of natural and technological hazards (NaTech events); nevertheless, existing policies frequently overlook the potential vulnerabilities of process plants to these complex phenomena. The goal of this research was to systematically characterize the vulnerability of industrial critical infrastructures (ICIs) to various hazards in their territories. A multi-scale procedure was implemented in the Italian context as a case study, where spatial analyses were developed using open data. Starting from the Italian national inventory, the MHIs were clustered in industrial macro-sectors and represented nationally by regions, relating their distribution to meteorological or geophysical data of interest. At the regional scale, the MHIs of the Piedmont Region were represented as punctual elements, associating the population within potential damage zones by province. At the municipal scale, a previously validated multi-hazard tool for vulnerability assessment was then tailored to a reduced scale for specific applications in an industrial context. This adaptation, which considers the two-way interaction between an energetic critical infrastructure and various hazards in its surroundings, delivers a spatial vulnerability profile that may complement the probabilistic analysis of industrial incidental scenarios. In summary, this framework may raise the stakeholders awareness at various levels and with different interests within the industrial accident control decision-making chain, from operators to competent authorities.
{"title":"Multi-scale characterization of industrial infrastructure vulnerability to multiple hazards in their territories","authors":"David Javier Castro Rodriguez, Antonello A. Barresi, Micaela Demichela","doi":"10.1016/j.jnlssr.2024.11.004","DOIUrl":"10.1016/j.jnlssr.2024.11.004","url":null,"abstract":"<div><div>Directive 2022/2557 from the European Commission aims to enhance the resilience of critical entities in Europe by integrating with existing European legislation, but it lacks explicit guidance on addressing vulnerabilities. Specifically, major hazard industries (MHIs) are critical infrastructures that face unique risks arising from the interactions of natural and technological hazards (NaTech events); nevertheless, existing policies frequently overlook the potential vulnerabilities of process plants to these complex phenomena. The goal of this research was to systematically characterize the vulnerability of industrial critical infrastructures (ICIs) to various hazards in their territories. A multi-scale procedure was implemented in the Italian context as a case study, where spatial analyses were developed using open data. Starting from the Italian national inventory, the MHIs were clustered in industrial macro-sectors and represented nationally by regions, relating their distribution to meteorological or geophysical data of interest. At the regional scale, the MHIs of the Piedmont Region were represented as punctual elements, associating the population within potential damage zones by province. At the municipal scale, a previously validated multi-hazard tool for vulnerability assessment was then tailored to a reduced scale for specific applications in an industrial context. This adaptation, which considers the two-way interaction between an energetic critical infrastructure and various hazards in its surroundings, delivers a spatial vulnerability profile that may complement the probabilistic analysis of industrial incidental scenarios. In summary, this framework may raise the stakeholders awareness at various levels and with different interests within the industrial accident control decision-making chain, from operators to competent authorities.</div></div>","PeriodicalId":62710,"journal":{"name":"安全科学与韧性(英文)","volume":"6 2","pages":"Pages 297-315"},"PeriodicalIF":3.7,"publicationDate":"2025-02-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143905904","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The COVID-19 pandemic has profoundly impacted startups, disrupting operations, consumer behavior, and market dynamics. Addressing these challenges necessitates an in-depth analysis of startups' vulnerabilities and the development of effective strategies to bolster their resilience and sustainability. This study introduces a combined thematic analysis and system dynamics approach to enhance startups' resilience during the pandemic. A qualitative thematic analysis was employed to identify the key factors influencing resilience. Semi-structured interviews with 12 experts provided data categorized into 21 themes across four dimensions: team, founder, human resources, and startup characteristics. Building on the qualitative phase, a system dynamics model was developed, comprising 32 auxiliary variables, five flow variables, four constants, and four stock variables. Four scenarios were devised to evaluate resilience within this model, reflecting varying degrees of financial strength, government support, and crisis management improvements. The results highlight the effectiveness of Scenario 4, which achieved the highest resilience improvement, driven by a 5 % increase in financial strength, a 5 % increase in government support, and a 10 % enhancement in crisis management. These findings offer critical insights for stakeholders and researchers seeking to strengthen startup resilience during crises.
{"title":"Modeling the resilience of startups in the COVID-19 pandemic using the system dynamics approach","authors":"Mahdi Homayounfar , Faezeh Kamali-Chirani , Adel Pourghader Chobar , Amir Daneshvar","doi":"10.1016/j.jnlssr.2024.10.004","DOIUrl":"10.1016/j.jnlssr.2024.10.004","url":null,"abstract":"<div><div>The COVID-19 pandemic has profoundly impacted startups, disrupting operations, consumer behavior, and market dynamics. Addressing these challenges necessitates an in-depth analysis of startups' vulnerabilities and the development of effective strategies to bolster their resilience and sustainability. This study introduces a combined thematic analysis and system dynamics approach to enhance startups' resilience during the pandemic. A qualitative thematic analysis was employed to identify the key factors influencing resilience. Semi-structured interviews with 12 experts provided data categorized into 21 themes across four dimensions: team, founder, human resources, and startup characteristics. Building on the qualitative phase, a system dynamics model was developed, comprising 32 auxiliary variables, five flow variables, four constants, and four stock variables. Four scenarios were devised to evaluate resilience within this model, reflecting varying degrees of financial strength, government support, and crisis management improvements. The results highlight the effectiveness of Scenario 4, which achieved the highest resilience improvement, driven by a 5 % increase in financial strength, a 5 % increase in government support, and a 10 % enhancement in crisis management. These findings offer critical insights for stakeholders and researchers seeking to strengthen startup resilience during crises.</div></div>","PeriodicalId":62710,"journal":{"name":"安全科学与韧性(英文)","volume":"6 3","pages":"Article 100185"},"PeriodicalIF":3.7,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144536048","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This study is novel, as it aims to generate an emergency scenario model for the analysis of dynamic risks in business parks to help decision-makers provide an optimal response in any emergency. To this end, the CIA-ISM methodology, which is the combination of Cross-Impact Analysis (CIA) and Interpretative Structural Model (ISM), allows the representation of all possible connections among risks, as well as representing real events under conditions of uncertainty. The proposed model integrates the use of an information system for the generation of multiple emergency scenarios that include the capture of complex interactions among agents, resources and variable environmental conditions. The results highlight the capacity of the proposed emergency scenario model based on CIA-ISM for the analysis of dynamic risks in business parks, identification of hidden vulnerabilities and evaluation of mitigation strategies in real-time. This study not only expands the theoretical knowledge of emergency management but also provides a useful tool to improve preparedness and response capacity in the face of adverse events in dynamic and complex environments.
{"title":"Emergency scenario modeling for the analysis of dynamic risks in business parks","authors":"Rodríguez Pillaga Renán Teodoro , Bañuls Víctor A．","doi":"10.1016/j.jnlssr.2024.11.002","DOIUrl":"10.1016/j.jnlssr.2024.11.002","url":null,"abstract":"<div><div>This study is novel, as it aims to generate an emergency scenario model for the analysis of dynamic risks in business parks to help decision-makers provide an optimal response in any emergency. To this end, the CIA-ISM methodology, which is the combination of Cross-Impact Analysis (CIA) and Interpretative Structural Model (ISM), allows the representation of all possible connections among risks, as well as representing real events under conditions of uncertainty. The proposed model integrates the use of an information system for the generation of multiple emergency scenarios that include the capture of complex interactions among agents, resources and variable environmental conditions. The results highlight the capacity of the proposed emergency scenario model based on CIA-ISM for the analysis of dynamic risks in business parks, identification of hidden vulnerabilities and evaluation of mitigation strategies in real-time. This study not only expands the theoretical knowledge of emergency management but also provides a useful tool to improve preparedness and response capacity in the face of adverse events in dynamic and complex environments.</div></div>","PeriodicalId":62710,"journal":{"name":"安全科学与韧性(英文)","volume":"6 2","pages":"Pages 265-279"},"PeriodicalIF":3.7,"publicationDate":"2025-01-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143860530","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-12DOI: 10.1016/j.jnlssr.2024.12.001
Ho Yin Wong , Meng Wang , Xiaoning Zhang , Yuxin Zhang , Ming Chi Wong , Xinyan Huang
Emergency exit sign systems guide occupants to safety, but they may fail in fires when smoke blocks routes. This study introduces an Intelligent Dynamic Exit Sign (IDES) system, integrating a fire-detection sensor network with dynamic sign patterns, which could enhance evacuation safety by always guiding occupants away from hazards. The system's operation framework and design rules ensure effective implementation. To address ethical concerns in complex scenarios, IDES includes a reversion mechanism that switches dynamic signs back to conventional static signs when necessary. The system's effectiveness is demonstrated through prototyping in a lab-scale tunnel model, assessing both the dynamic patterns and hardware reliability. Results show the potential of IDES to automatically optimize evacuation procedures and occupant safety during emergencies. Furthermore, the study delves into challenges associated with real-world implementation and offers insights for future applications of this innovative safety solution in more complex built environments.
{"title":"Safe evacuation framework with intelligent dynamic exit sign system and demonstration in tunnel fire","authors":"Ho Yin Wong , Meng Wang , Xiaoning Zhang , Yuxin Zhang , Ming Chi Wong , Xinyan Huang","doi":"10.1016/j.jnlssr.2024.12.001","DOIUrl":"10.1016/j.jnlssr.2024.12.001","url":null,"abstract":"<div><div>Emergency exit sign systems guide occupants to safety, but they may fail in fires when smoke blocks routes. This study introduces an Intelligent Dynamic Exit Sign (IDES) system, integrating a fire-detection sensor network with dynamic sign patterns, which could enhance evacuation safety by always guiding occupants away from hazards. The system's operation framework and design rules ensure effective implementation. To address ethical concerns in complex scenarios, IDES includes a reversion mechanism that switches dynamic signs back to conventional static signs when necessary. The system's effectiveness is demonstrated through prototyping in a lab-scale tunnel model, assessing both the dynamic patterns and hardware reliability. Results show the potential of IDES to automatically optimize evacuation procedures and occupant safety during emergencies. Furthermore, the study delves into challenges associated with real-world implementation and offers insights for future applications of this innovative safety solution in more complex built environments.</div></div>","PeriodicalId":62710,"journal":{"name":"安全科学与韧性(英文)","volume":"6 3","pages":"Article 100183"},"PeriodicalIF":3.7,"publicationDate":"2025-01-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144322423","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-31DOI: 10.1016/j.jnlssr.2024.11.001
Ran Li , Xiaofei Ye , Shuyi Pei , Xingchen Yan , Tao Wang , Jun Chen , Pengjun Zheng
In the context of the COVID-19 epidemic, a “double-hazard scenario” consisting of a natural disaster and a public health event simultaneously occurring is more likely to arise. However, compared with single-hazard, multiple disasters confront the challenges of complexity, diversity, and demand urgency. To improve the efficiency of emergency material distribution under multiple disasters, this study first divided multiple disasters into three categories: independent scenario, sequential scenario, and coupling scenario. A set of evaluation index systems for multiple disasters was established to quantify the urgency of demand. The routing optimization model of emergency vehicles for multiple disasters was proposed by combining demand urgency and road damage, and the non-dominated sorting genetic algorithm II (NSGA-II) was used to simulate and validate the model. A coupling scenario considering two typical disasters of hurricanes and epidemics was selected as a validation example, and sensitivity analysis was also performed for different algorithms, scenarios, and constraints. The results demonstrated that the proposed model could effectively address the vehicle routing problem of emergency materials in the context of multiple disasters. Compared to the NSGA, the NSGA-II was used to reduce the total delivery time, cost, and penalty cost by 15.98%, 13.60%, and 16.14%, respectively. Compared with the independent scenario, the coupling scenario increased the total delivery time and cost by 186.28% and 132.48% during the epidemic. However, it reduced the total delivery time by 4.00% and increased the delivery cost by 23.55% compared with the hurricane. Compared with the model without consideration, the model considering demand urgency and road damage reduced the total delivery time and cost by 17.88% and 8.73%, respectively. The model constructed in this study addressed the vehicle routing problem considering the demand urgency and road damage in the optimization process, particularly in the context of multiple disasters.
{"title":"Optimization of vehicle routing problems combining the demand urgency and road damage for multiple disasters","authors":"Ran Li , Xiaofei Ye , Shuyi Pei , Xingchen Yan , Tao Wang , Jun Chen , Pengjun Zheng","doi":"10.1016/j.jnlssr.2024.11.001","DOIUrl":"10.1016/j.jnlssr.2024.11.001","url":null,"abstract":"<div><div>In the context of the COVID-19 epidemic, a “double-hazard scenario” consisting of a natural disaster and a public health event simultaneously occurring is more likely to arise. However, compared with single-hazard, multiple disasters confront the challenges of complexity, diversity, and demand urgency. To improve the efficiency of emergency material distribution under multiple disasters, this study first divided multiple disasters into three categories: independent scenario, sequential scenario, and coupling scenario. A set of evaluation index systems for multiple disasters was established to quantify the urgency of demand. The routing optimization model of emergency vehicles for multiple disasters was proposed by combining demand urgency and road damage, and the non-dominated sorting genetic algorithm II (NSGA-II) was used to simulate and validate the model. A coupling scenario considering two typical disasters of hurricanes and epidemics was selected as a validation example, and sensitivity analysis was also performed for different algorithms, scenarios, and constraints. The results demonstrated that the proposed model could effectively address the vehicle routing problem of emergency materials in the context of multiple disasters. Compared to the NSGA, the NSGA-II was used to reduce the total delivery time, cost, and penalty cost by 15.98%, 13.60%, and 16.14%, respectively. Compared with the independent scenario, the coupling scenario increased the total delivery time and cost by 186.28% and 132.48% during the epidemic. However, it reduced the total delivery time by 4.00% and increased the delivery cost by 23.55% compared with the hurricane. Compared with the model without consideration, the model considering demand urgency and road damage reduced the total delivery time and cost by 17.88% and 8.73%, respectively. The model constructed in this study addressed the vehicle routing problem considering the demand urgency and road damage in the optimization process, particularly in the context of multiple disasters.</div></div>","PeriodicalId":62710,"journal":{"name":"安全科学与韧性(英文)","volume":"6 2","pages":"Pages 196-211"},"PeriodicalIF":3.7,"publicationDate":"2024-12-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143829389","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The primary goal of this research study is to enhance energy resilience with a focus on cost efficiency. To achieve this objective, two key objectives have been identified: (1) reducing unserved loads, and (2) implementing cost-effective resource allocation strategies. A high-fidelity detailed model of a solar plus storage microgrid is developed to simulate a variety of what-if scenarios. This model is based on the conceptual design of a campus microgrid facility, which is slated for commissioning at UL Lafayette in close collaboration with a local power utility. The study examines the microgrid’s performance under different configurations, including both stationary battery and mobile battery storage options. To ensure the realism of the scenarios, real solar data from specific days following the occurrence of three major hurricanes in Louisiana is utilized. The analysis includes an assessment of unserved loads under various scenarios, as well as an investigation into the resilience impact of investment decisions and the planning and operation of mobile storage systems. The results indicate the proposed planning and operation will improve resilience while staying within the profitable range. The resilience is quantified and compared with other scenarios providing an insightful planning framework for decision-makers.
{"title":"Cost-aware strategies for enhancing energy resilience in microgrids via stationary and mobile resources","authors":"S.M. Safayet Ullah , Kouhyar Sheida , Farzad Ferdowsi , Terrence Chambers","doi":"10.1016/j.jnlssr.2024.10.002","DOIUrl":"10.1016/j.jnlssr.2024.10.002","url":null,"abstract":"<div><div>The primary goal of this research study is to enhance energy resilience with a focus on cost efficiency. To achieve this objective, two key objectives have been identified: (1) reducing unserved loads, and (2) implementing cost-effective resource allocation strategies. A high-fidelity detailed model of a solar plus storage microgrid is developed to simulate a variety of what-if scenarios. This model is based on the conceptual design of a campus microgrid facility, which is slated for commissioning at UL Lafayette in close collaboration with a local power utility. The study examines the microgrid’s performance under different configurations, including both stationary battery and mobile battery storage options. To ensure the realism of the scenarios, real solar data from specific days following the occurrence of three major hurricanes in Louisiana is utilized. The analysis includes an assessment of unserved loads under various scenarios, as well as an investigation into the resilience impact of investment decisions and the planning and operation of mobile storage systems. The results indicate the proposed planning and operation will improve resilience while staying within the profitable range. The resilience is quantified and compared with other scenarios providing an insightful planning framework for decision-makers.</div></div>","PeriodicalId":62710,"journal":{"name":"安全科学与韧性(英文)","volume":"6 2","pages":"Pages 163-174"},"PeriodicalIF":3.7,"publicationDate":"2024-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143792727","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-28DOI: 10.1016/j.jnlssr.2024.09.002
Hafiz Mughees Ahmad, Afshin Rahimi
Workplace accidents continue to pose significant human safety risks, particularly in the construction and manufacturing industries. The necessity for effective Personal Protective Equipment (PPE) compliance has become increasingly paramount. We focus on developing non-invasive techniques based on the Object Detection (OD) and Convolutional Neural Network (CNN). The aim is to detect and verify the proper use of various types of PPE such as helmets, safety glasses, masks, and protective clothing. This study proposes the SH17 Dataset, consisting of 8,099 annotated images containing 75,994 instances of 17 classes collected from diverse industrial environments, to train and validate the OD models. We have trained state-of-the-art OD models for benchmarking, and initial results demonstrate promising accuracy levels with You Only Look Once (YOLO)v9-e model variant exceeding 70.9% in PPE detection. The validation of the model across cross-domain datasets indicates that integrating these technologies can substantially enhance safety management systems. This approach offers a scalable and efficient solution for industries seeking to comply with human safety regulations while safeguarding their workforce. The dataset is available at https://github.com/ahmadmughees/sh17dataset.
工作场所意外持续构成重大的人身安全风险，特别是在建造业和制造业。有效遵守个人防护装备（PPE）的必要性变得越来越重要。我们专注于开发基于目标检测（OD）和卷积神经网络（CNN）的非侵入性技术。目的是检测和核实正确使用各种类型的个人防护装备，如头盔、安全眼镜、口罩和防护服。本研究提出了SH17数据集，该数据集由8099张带注释的图像组成，包含从不同工业环境中收集的17个类别的75,994个实例，用于训练和验证OD模型。我们已经训练了最先进的OD模型进行基准测试，初步结果表明，You Only Look Once (YOLO)v9-e模型变体在PPE检测方面的准确率超过70.9%。跨领域数据集的模型验证表明，集成这些技术可以大大增强安全管理系统。此方法为寻求在保护其劳动力的同时遵守人类安全法规的行业提供了可扩展且高效的解决方案。该数据集可在https://github.com/ahmadmughees/sh17dataset上获得。
{"title":"SH17: A dataset for human safety and personal protective equipment detection in manufacturing industry","authors":"Hafiz Mughees Ahmad, Afshin Rahimi","doi":"10.1016/j.jnlssr.2024.09.002","DOIUrl":"10.1016/j.jnlssr.2024.09.002","url":null,"abstract":"<div><div>Workplace accidents continue to pose significant human safety risks, particularly in the construction and manufacturing industries. The necessity for effective Personal Protective Equipment (PPE) compliance has become increasingly paramount. We focus on developing non-invasive techniques based on the Object Detection (OD) and Convolutional Neural Network (CNN). The aim is to detect and verify the proper use of various types of PPE such as helmets, safety glasses, masks, and protective clothing. This study proposes the SH17 Dataset, consisting of 8,099 annotated images containing 75,994 instances of 17 classes collected from diverse industrial environments, to train and validate the OD models. We have trained state-of-the-art OD models for benchmarking, and initial results demonstrate promising accuracy levels with You Only Look Once (YOLO)v9-e model variant exceeding 70.9% in PPE detection. The validation of the model across cross-domain datasets indicates that integrating these technologies can substantially enhance safety management systems. This approach offers a scalable and efficient solution for industries seeking to comply with human safety regulations while safeguarding their workforce. The dataset is available at <span><span>https://github.com/ahmadmughees/sh17dataset</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":62710,"journal":{"name":"安全科学与韧性(英文)","volume":"6 2","pages":"Pages 175-185"},"PeriodicalIF":3.7,"publicationDate":"2024-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143807336","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-22DOI: 10.1016/j.jnlssr.2024.10.001
Chupei Liao, Kuoyi Lin
Driver behavior is a critical factor in road safety, highlighting the need for advanced methods in Distracted Driving Classification (DDC). In this study, we introduce DDC-Chat, a novel classification method based on a Visual large Language Model (VLM). DDC-Chat is an interactive multimodal system built upon LLAVA-Plus, fine-tuned specifically for addressing distracted driving detection. It utilizes logical reasoning chains to activate visual skills, including segmentation and pose detection, through end-to-end training. Furthermore, instruction tuning allows DDC-Chat to continuously incorporate new visual skills, enhancing its ability to classify distracted driving behavior. Our extensive experiments demonstrate that DDC-Chat achieves state-of-the-art performance on public DDC datasets, surpassing previous benchmarks. In evaluations on the 100-Driver dataset, the model exhibits superior results in both zero-shot and few-shot learning contexts, establishing it as a valuable tool for improving driving safety by accurately identifying driver distraction. Due to the computational intensity of inference, DDC-Chat is optimized for deployment on remote servers, with data streamed from in-vehicle monitoring systems for real-time analysis.
{"title":"DDC-Chat: Achieving accurate distracted driver classification through instruction tuning of visual language model","authors":"Chupei Liao, Kuoyi Lin","doi":"10.1016/j.jnlssr.2024.10.001","DOIUrl":"10.1016/j.jnlssr.2024.10.001","url":null,"abstract":"<div><div>Driver behavior is a critical factor in road safety, highlighting the need for advanced methods in <strong>D</strong>istracted <strong>D</strong>riving <strong>C</strong>lassification (DDC). In this study, we introduce DDC-Chat, a novel classification method based on a <strong>V</strong>isual large <strong>L</strong>anguage <strong>M</strong>odel (VLM). DDC-Chat is an interactive multimodal system built upon LLAVA-Plus, fine-tuned specifically for addressing distracted driving detection. It utilizes logical reasoning chains to activate visual skills, including segmentation and pose detection, through end-to-end training. Furthermore, instruction tuning allows DDC-Chat to continuously incorporate new visual skills, enhancing its ability to classify distracted driving behavior. Our extensive experiments demonstrate that DDC-Chat achieves state-of-the-art performance on public DDC datasets, surpassing previous benchmarks. In evaluations on the 100-Driver dataset, the model exhibits superior results in both zero-shot and few-shot learning contexts, establishing it as a valuable tool for improving driving safety by accurately identifying driver distraction. Due to the computational intensity of inference, DDC-Chat is optimized for deployment on remote servers, with data streamed from in-vehicle monitoring systems for real-time analysis.</div></div>","PeriodicalId":62710,"journal":{"name":"安全科学与韧性(英文)","volume":"6 2","pages":"Pages 250-264"},"PeriodicalIF":3.7,"publicationDate":"2024-11-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143860683","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-18DOI: 10.1016/j.jnlssr.2024.08.002
Muammer Yaman , Cüneyt Kurtay
Fires that occur in assembly buildings cause great loss of life and property. Children's cultural centers included in assembly buildings should also be evaluated within this context. Children create an occupant profile in cultural centers, and the internal environment has an excessive fire load, which poses a great risk of fires. In fire evacuation scenarios for children's cultural centers, it is necessary to create appropriate evacuation conditions. In this work, fire safety was analyzed over total evacuation time within occupant-based fire evacuation simulations of a children's cultural center located in Istanbul. The effect of the children's theatre hall located on the top floor on a building's total evacuation time has been studied. The effectiveness of alternative fire escape routes on evacuation time through different evacuation scenarios has been analyzed, and safe evacuation strategies for children's cultural centers have been revealed. As a result of this study, recommendations were presented within performance-based fire evacuation strategies in the design of children's cultural centers. As the future of the countries, fire safety design criteria have been created in children's cultural centers for children to be able to be safe in educational, cultural, and artistic environments.
{"title":"Analysis of fire evacuation scenarios in children's cultural centers","authors":"Muammer Yaman , Cüneyt Kurtay","doi":"10.1016/j.jnlssr.2024.08.002","DOIUrl":"10.1016/j.jnlssr.2024.08.002","url":null,"abstract":"<div><div>Fires that occur in assembly buildings cause great loss of life and property. Children's cultural centers included in assembly buildings should also be evaluated within this context. Children create an occupant profile in cultural centers, and the internal environment has an excessive fire load, which poses a great risk of fires. In fire evacuation scenarios for children's cultural centers, it is necessary to create appropriate evacuation conditions. In this work, fire safety was analyzed over total evacuation time within occupant-based fire evacuation simulations of a children's cultural center located in Istanbul. The effect of the children's theatre hall located on the top floor on a building's total evacuation time has been studied. The effectiveness of alternative fire escape routes on evacuation time through different evacuation scenarios has been analyzed, and safe evacuation strategies for children's cultural centers have been revealed. As a result of this study, recommendations were presented within performance-based fire evacuation strategies in the design of children's cultural centers. As the future of the countries, fire safety design criteria have been created in children's cultural centers for children to be able to be safe in educational, cultural, and artistic environments.</div></div>","PeriodicalId":62710,"journal":{"name":"安全科学与韧性(英文)","volume":"6 1","pages":"Pages 114-123"},"PeriodicalIF":3.7,"publicationDate":"2024-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143388318","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号