Toward an Integrated Disaster Management Approach: How Artificial Intelligence Can Boost Disaster Management

Sustainability, Journal Year: 2021, Volume and Issue: 13(22), P. 12560 - 12560

Published: Nov. 13, 2021

Technical and methodological enhancement of hazards disaster research is identified as a critical question in management. Artificial intelligence (AI) applications, such tracking mapping, geospatial analysis, remote sensing techniques, robotics, drone technology, machine learning, telecom network services, accident hot spot smart city urban planning, transportation environmental impact are the technological components societal change, having significant implications for on response to disasters. Social science researchers have used various technologies methods examine disasters through disciplinary, multidisciplinary, interdisciplinary lenses. They employed both quantitative qualitative data collection analysis strategies. This study provides an overview current applications AI management during its four phases how vital all phases, leading faster, more concise, equipped response. Integrating geographic information system (GIS) (RS) into enables higher situational awareness, recovery operations. GIS RS commonly recognized key support tools Visualization capabilities, satellite images, artificial can assist governments making quick decisions after natural

Language: Английский

---

The State of the Art in Deep Learning Applications, Challenges, and Future Prospects: A Comprehensive Review of Flood Forecasting and Management

Sustainability, Journal Year: 2023, Volume and Issue: 15(13), P. 10543 - 10543

Published: July 4, 2023

Floods are a devastating natural calamity that may seriously harm both infrastructure and people. Accurate flood forecasts control essential to lessen these effects safeguard populations. By utilizing its capacity handle massive amounts of data provide accurate forecasts, deep learning has emerged as potent tool for improving prediction control. The current state applications in forecasting management is thoroughly reviewed this work. review discusses variety subjects, such the sources utilized, models used, assessment measures adopted judge their efficacy. It assesses approaches critically points out advantages disadvantages. article also examines challenges with accessibility, interpretability models, ethical considerations prediction. report describes potential directions deep-learning research enhance predictions Incorporating uncertainty estimates into integrating many sources, developing hybrid mix other methodologies, enhancing few these. These goals can help become more precise effective, which will result better plans forecasts. Overall, useful resource academics professionals working on topic management. reviewing art, emphasizing difficulties, outlining areas future study, it lays solid basis. Communities prepare destructive floods by implementing cutting-edge algorithms, thereby protecting people infrastructure.

Language: Английский

---

Water body classification from high-resolution optical remote sensing imagery: Achievements and perspectives

ISPRS Journal of Photogrammetry and Remote Sensing, Journal Year: 2022, Volume and Issue: 187, P. 306 - 327

Published: March 25, 2022

Language: Английский

---

Machine Learning-Aided Operations and Communications of Unmanned Aerial Vehicles: A Contemporary Survey

IEEE Communications Surveys & Tutorials, Journal Year: 2023, Volume and Issue: 26(1), P. 496 - 533

Published: Sept. 11, 2023

Over the past decade, Unmanned Aerial Vehicles (UAVs) have provided pervasive, efficient, and cost-effective solutions for data collection communications. Their excellent mobility, flexibility, fast deployment enable UAVs to be extensively utilized in agriculture, medical, rescue missions, smart cities, intelligent transportation systems. Machine learning (ML) has been increasingly demonstrating its capability of improving automation operation precision many UAV-assisted applications, such as communications, sensing, collection. The ongoing amalgamation UAV ML techniques is creating a significant synergy empowering with unprecedented intelligence autonomy. This survey aims provide timely comprehensive overview used operations communications identify potential growth areas research gaps. We emphasize four key components which can significantly contribute, namely, perception feature extraction, interpretation regeneration, trajectory mission planning, aerodynamic control operation. classify latest popular tools based on their applications conduct gap analyses. also takes step forward by pointing out challenges upcoming realm ML-aided automated It revealed that different dominate modules While there an increasing trend cross-module designs, little effort devoted end-to-end framework, from extraction unveiled reliability trust require attention before full cooperation between humans come fruition.

Language: Английский

---

An Efficient Smart Flood Detection and Alert System based on Automatic Water Level Recorder Approach using IoT

International Journal of Computational and Experimental Science and Engineering, Journal Year: 2025, Volume and Issue: 11(1)

Published: Jan. 12, 2025

An innovative flood detection system may track an increase in water levels. Deployed cities or other areas of interest, the consists sensors. Both mains energy and solar power are viable options for To detect impending flooding promptly, a warning uses reliable up-to-date sensing equipment such as rain gauges, level sensors, flow rate sensors smart alert system. The challenging characteristics systems that some people not be able to access warnings, flash floods happen too quickly adequate. Hence, proposed method, Automatic Water Level Recorder enabled Internet Things (AWLR-IoT), which integrates low-cost cloud overcome challenges optimization modelling efficiency. Among most destructive natural catastrophes on Earth is flooding. After that, Wireless Sensor Network (WSN) used accomplish prediction utilizing data from by Things. Heavy rainfall following outflow cause nations with certain climate conditions. monitors humidity, temperature, level, rise rate, identify when imminent. function AWLR-IoT sensor monitoring recording database real-time sensing. This research shows has reduced processing time compared conventional processing.

Language: Английский

---

Inspecting Buildings Using Drones and Computer Vision: A Machine Learning Approach to Detect Cracks and Damages

Drones, Journal Year: 2021, Volume and Issue: 6(1), P. 5 - 5

Published: Dec. 24, 2021

Manual inspection of infrastructure damages such as building cracks is difficult due to the objectivity and reliability assessment high demands time costs. This can be automated using unmanned aerial vehicles (UAVs) for imagery damages. Numerous computer vision-based approaches have been applied address limitations crack detection but they their that overcome by various hybrid based on artificial intelligence (AI) machine learning (ML) techniques. The convolutional neural networks (CNNs), an application deep (DL) method, display remarkable potential automatically detecting image features are less sensitive noise. A modified hierarchical CNN architecture has used in this study damage civil infrastructures. proposed 16 convolution layers a cycle generative adversarial network (CycleGAN). For study, images were collected UAVs open-source mid rise buildings (five stories above) constructed during 2000 Sydney, Australia. Conventionally, only utilizes last layer convolution. However, our utility multiple layers. Another important component guided filtering (GF) conditional random fields (CRFs) refine predicted outputs get reliable results. Benchmarking data (600 images) Sydney-based was test architecture. produced superior performance when evaluated five methods: GF Baseline (BN) Deep-Crack BN, GF, SegNet. Overall, method outperformed all other methods indicated global accuracy (0.990), class average (0.939), mean intersection union overall classes (IoU) (0.879), precision (0.838), recall F-score (0.8581) values. provides advantages reduced noise, highly integrated supervision features, adequate learning, aggregation both multi-scale multilevel training procedure along with refinement output predictions.

Language: Английский

---

Automatic Target Detection from Satellite Imagery Using Machine Learning

Sensors, Journal Year: 2022, Volume and Issue: 22(3), P. 1147 - 1147

Published: Feb. 2, 2022

Object detection is a vital step in satellite imagery-based computer vision applications such as precision agriculture, urban planning and defense applications. In imagery, object very complicated task due to various reasons including low pixel resolution of objects small the large scale (a single image taken by Digital Globe comprises over 240 million pixels) images. images has many challenges class variations, multiple pose, high variance size, illumination dense background. This study aims compare performance existing deep learning algorithms for imagery. We created dataset imagery perform using convolutional neural network-based frameworks faster RCNN (faster region-based network), YOLO (you only look once), SSD (single-shot detector) SIMRDWN (satellite multiscale rapid with windowed networks). addition that, we also performed an analysis these approaches terms accuracy speed developed The results showed that 97% on high-resolution images, while Faster 95.31% standard (1000 × 600). YOLOv3 94.20% (416 416) other hand 84.61% (300 300). When it comes efficiency, obvious leader. real-time surveillance, fails. takes 170 190 milliseconds task, 5 103 milliseconds.

Language: Английский

---

Environmentally-Aware and Energy-Efficient Multi-Drone Coordination and Networking for Disaster Response

IEEE Transactions on Network and Service Management, Journal Year: 2023, Volume and Issue: 20(2), P. 1093 - 1109

Published: Feb. 10, 2023

In a disaster response management (DRM) scenario, communication and coordination are limited, absence of related infrastructure hinders situational awareness. Unmanned aerial vehicles (UAVs) or drones provide new capabilities for DRM to address these barriers. However, there is dearth works that multiple heterogeneous collaboratively working together form flying ad-hoc network (FANET) with air-to-air air-to-ground links impacted by: (i) environmental obstacles, (ii) wind, (iii) limited battery capacities. this paper, we present novel environmentally-aware energy-efficient multi-drone networking scheme features Reinforcement Learning (RL) based location prediction algorithm coupled packet forwarding drone-to-ground establishment. We specifically two drone location-based solutions (i.e., heuristic greedy, learning-based) in our approach support application requirements. These requirements involve improving connectivity optimize delivery ratio end-to-end delay) despite efficiency by lower energy use time consumption) constraints. evaluate state-of-the-art algorithms trace-based FANET simulation testbed featuring rural metropolitan areas. Results show strategy overcomes obstacles can achieve 81-to-90% performance observed under no obstacle conditions. the presence improves 14-to-38% while also providing 23-to-54% savings areas; same areas achieved an average 25% gain when compared baseline awareness approaches 15-to-76% savings.

Language: Английский

---

Vehicle Detection and Classification via YOLOv8 and Deep Belief Network over Aerial Image Sequences

Sustainability, Journal Year: 2023, Volume and Issue: 15(19), P. 14597 - 14597

Published: Oct. 8, 2023

Vehicle detection and classification are the most significant challenging activities of an intelligent traffic monitoring system. Traditional methods highly computationally expensive also impose restrictions when mode data collection changes. This research proposes a new approach for vehicle over aerial image sequences. The proposed model consists five stages. All images preprocessed in first stage to reduce noise raise brightness level. foreground items then extracted from these using segmentation. segmented passed onto YOLOv8 algorithm detect locate vehicles each image. feature extraction phase is applied detected vehicles. involves Scale Invariant Feature Transform (SIFT), Oriented FAST Rotated BRIEF (ORB), KAZE features. For classification, we used Deep Belief Network (DBN) classifier. Based on experimental results across three datasets produced better outcomes; attained accuracy 95.6% Detection Aerial Imagery (VEDAI) 94.6% Drone (VAID) dataset, respectively. To compare our with other standard techniques, have drawn comparative analysis latest techniques research.

Language: Английский

---

Efficient CNN-based disaster events classification using UAV-aided images for emergency response application

Neural Computing and Applications, Journal Year: 2024, Volume and Issue: 36(18), P. 10599 - 10612

Published: March 27, 2024

Language: Английский

---