Situational Awareness Research Articles

The Impact of Constant Observation in Pediatric Mental Health Emergencies: Perspectives from Emergency Department Staff

ABSTRACT Background Emergency departments are critical care settings for children in mental and behavioral health crises. Psychiatrists and psychiatric-mental health nurses collaborate with emergency department teams to manage safety risks for these patients, which often involves constant patient observation, particularly for agitated or suicidal patients. Objective This paper aims to understand emergency department staff’s perspectives on constant observation of pediatric mental and behavioral health patients and the impact of this practice on emergency department workflows. Methods This qualitative study thematically analyzes 55 semi-structured interviews with members of the emergency department team, including patient observers, from four different hospitals. Results Eight themes were identified as impacting emergency department workflows and patient care. The most commonly mentioned themes included the impact of constant observation on emergency department workflow, situational awareness and clear sightlines to the patient, and compliance with constant observation-related organizational policies. Conclusions Findings provide insights for optimizing the role of the patient observer, incorporating remote monitoring to increase patient safety, enhancing the emergency department physical environment to enhance the quality of care, optimizing patient outcomes, and maintaining a safe environment. This study’s results support advanced psychiatric-mental health practice nurses and emergency and psychiatric-mental health clinical nurses in improving the quality of care and safety for this vulnerable population.

Implementing Doppler Radar Using ADAS

1. Abstract This document examines the role of Doppler Radar technology in Advanced Driver Assistance Systems (ADAS) and highlights its significance in enhancing vehicle safety and enabling autonomous functions, marking a crucial advancement towards safer, more self-reliant vehicles. By exploring the basic concepts of Doppler Radar and its applications in detecting objects, measuring speed, and preventing collisions, we demonstrate how radar contributes to better situational awareness in ever-changing traffic scenarios. This paper delves into key aspects of Doppler Radar, the architectural framework of the system, signal processing methods, and the collaborative benefits achieved through its integration with various other sensing technologies via sensor fusion. It also addresses current challenges such as resolution, environmental factors, and regulatory issues, concluding with perspectives on future developments in high-resolution imaging radar and sensor networks powered by artificial intelligence for autonomous driving. 2. Keywords Doppler radar, Advanced Driver Assistance Systems (ADAS), autonomous vehicles, Doppler effect, radar signal processing, adaptive cruise control (ACC), forward collision warning (FCW), automatic emergency braking (AEB), blind spot detection, rear cross traffic alert, traffic sign recognition, radar frequency bands, LiDAR, camera-based detection, environmental interference, multipath reflection, angular resolution, 4D imaging radar, electronic control unit (ECU), spectrum management.

Enhancing Rescue Operations in Dubai Police’s Transport and Rescue Department through IoT and AI Integration

Purpose: This research proposes how all the space, the Internet of Things, and Artificial Intelligence technologies can be better used in the Transport and Rescue Department of the Dubai Police. With the rapid evolution in smart technologies, IoT and AI have much scope to refine rescue operations regarding time management, accuracy, resources, and overall processes. Since the requirements in rescue missions are often complex, including environmental change and the need for urgency, the role of IoT and AI is extremely important in decision-making and successful operation. This study investigates how to employ these technologies within Dubai Police's existing rescue framework, which has gaps such as slow response or response delay and inefficient resource distribution. It aims to explore the potential benefits and challenges that the use of these technologies entails while drawing lessons from global best practices on emergency response more generally. Methodology: Using qualitative and quantitative research methods, including interviews, surveys, observation, global case studies, and reviews of existing literature, the research will make recommendations on how these technologies can be effectively used to enhance the success of rescue missions. By fully embracing the concept of Dubai being a smart city, this research will not only upgrade the local emergency response systems, but it will also create a model for the other regions that want to convert to advanced technologies in rescue operations. At the heart of the research is a goal to reach more efficient, faster, and safer rescue operations that, in turn, ensure public safety in Dubai. Findings: Research results show that the Dubai Police have already started using a variety of technology to ensure smart operations (e.g., SAS analytics and VR training), but IoT and AI integration are hampered by infrastructure, technology training gaps, and interagency coordination issues. Rescuers recognize that AI and IoT can be utilized to help in enhancing situational awareness and speeding the response, however, they also flag issues about cybersecurity and the system's compatibility. Unique Contribution to Theory, Practice and Policy: The implication of the research is two-fold. First, it gives specific, practical recommendations on what to do to enhance Dubai's emergency response system. Second, it presents a scalable model for cities that are trying to equip themselves with smart, technology-driven rescue systems.

Social sensing a volcanic eruption: application to Kīlauea, 2018

Abstract. Protecting lives and livelihoods during volcanic eruptions is the key challenge in volcanology, conducted primarily by volcano monitoring and emergency management organisations, but it is complicated by scarce knowledge of how communities respond in times of crisis. Social sensing is a rapidly developing practice that can be adapted for volcanology. Here we use social sensing of Twitter (currently known as X) posts to track changes in social action and reaction throughout the 2018 eruption of Kīlauea on the island of Hawai`i. The volume of relevant posts very rapidly increases in early May, coincident with the beginning of the eruption; automated sentiment analysis shows a simultaneous shift towards more negative emotions being expressed in post text. Substantial negative trends in sentiment are evident in reaction to high-impact events, including the destruction of a popular residential area and injuries sustained by tourists viewing the eruption. Topics of local Twitter conversation reveal societal actions, including the sharing of hazard warnings, mitigation actions, and aid announcements. Temporal trends in societal actions reflect patterns in volcanic activity (e.g. the peak and waning of eruptive activity), civil protection actions (e.g. risk mitigation actions and the communication of official warnings), and socioeconomic pressures (e.g. the destruction of homes). Local tweets detailing eruption damage and disruption display a similar temporal trend to independent estimates of the number of buildings in contact with lava. We show how hazard and risk information is discussed and reacted to on Twitter, which helps inform our understanding of community response actions and aids situational awareness, and outline how our approach could be adapted for use in real time.

AI-Driven Real-Time Surveillance: Anomaly Detection and Notification System

ABSTRACT Surveillance technologies are rapidly transitioning from traditional passive monitoring systems to intelligent platforms capable of detecting anomalies in real-time. This study presents an AI-powered surveillance framework designed to identify violent behavior across diverse input formats, such as static images, recorded videos, and live webcam feeds. Leveraging advanced deep learning architectures, including YOLO and TensorFlow-based models, the system delivers high-precision violence detection with minimal latency. On identifying a threat, it initiates instant alerts through visual prompts and sound notifications, ensuring rapid situational awareness. The interface is intuitively built to support real-time configuration of detection parameters and the monitoring of performance indicators such as detection frequency, operational uptime, and frame processing speed. Engineered for scalability and resilience, the system demonstrates strong applicability in domains like public safety, institutional monitoring, and content regulation. Its core advantages include high detection accuracy, efficient real-time processing, and adaptability to various data sources. By combining cutting edge AI strategies with responsive caution instruments, the framework offers a reliable, computerized arrangement for improving danger location in observation situations. Keywords: Real-Time Surveillance, Anomaly Detection, Violence Detection, YOLO, TensorFlow, Deep Learning, Smart Monitoring, Alert System

USE OF MODERN TECHNOLOGIES (VIRTUAL REALITY, SIMULATORS) FOR MODELING COMBAT CONDITIONS DURING PHYSICAL TRAINING

This article investigates the essential role of integrating physical and psychological training in preparing military personnel to face the demands of contemporary warfare. Modern military conflicts, such as the ongoing russian-Ukrainian war, emphasize the necessity for a holistic approach that combines physical fitness with psychological resilience to enhance operational effectiveness under extreme conditions. The study analyzes various methods, including functional fitness programs, combat-specific simulations, and stress-inducing training scenarios, to develop both physical and cognitive skills required for modern combat. Functional fitness, characterized by exercises replicating real-life combat movements, not only improves physical endurance but also strengthens situational awareness and decision-making. Similarly, psychological preparation focuses on developing adaptability, composure under stress, and rapid decision-making skills, which are critical during high-pressure situations.The research highlights the importance of aligning training programs with the specific needs of military roles, emphasizing exercises that simulate the challenges faced in combat. It also explores the role of advanced technologies such as virtual reality (VR) and simulation systems in creating immersive environments that prepare personnel for unpredictable battlefield scenarios. For instance, VR simulations allow trainees to experience realistic combat conditions, enhancing both individual performance and team coordination.The study discusses the psychological impact of war on soldiers and emphasizes that physical training should not only focus on endurance and strength but also on fostering psychological resilience. By integrating stress-inducing scenarios into physical training, personnel can build resilience and learn to make critical decisions in chaotic environments. The findings underscore the value of this approach in preparing soldiers for the physical and mental demands of extended deployments and active combat.

A study of the relationship between EEG and situation awareness of forklift drivers in task scenarios with different complexities

Situation awareness (SA) loss in forklift operations is a major human error source. Accurately assessing the SA levels of forklift drivers is essential. While physiological sensors such as electroencephalography (EEG) have been proposed to measure SA, the EEG–SA relationship remains unclear. This study analyzed EEG data from 11 forklift operators, comparing EEG states between high and low SA groups. Using permutation tests, the study assessed multi-band EEG correlations (F, C, P, O regions) with SA and examined how scenario complexity influenced these relationships. Results revealed distinct EEG patterns (θ, α and β waves in F, C, P and O regions) between high and low SA groups: θ / β, θ / (α + β) in the C and P regions, and θ / β, (α + θ) / (α + β) in the O region consistently correlating with SA across scenarios – and these indicators strengthened with scenario complexity. The findings advance EEG-based SA assessment, aiding accurate SA measurement in real-world settings.

Anomaly Detection in CCTV Videos Using LSTM Architecture

Abstract: The need for effective, automated anomaly detection systems that can handle enormous volumes of video data has been highlighted by the sharp rise in CCTV surveillance in both public and private areas. In order to improve situational awareness and give security personnel useful intelligence, this study proposes a model based on deep learning for identifying and measuring anomalies in CCTV data. Utilizing computer vision, the Convolutional neural networks (CNNs) are used in the model to extract spatial information. which allows it to identify and categorize anomalous occurrences in a variety of contexts. Furthermore, Long Short-Term Memory (LSTM) networks or recurrent neural networks (RNNs) are utilized to investigate mobility and activity patterns across time. This enhances the model's comprehension of sequence-based activities and enables it to identify deviations across time. Transfer learning techniques are used to improve performance even more, enabling the model to adapt effectively in a variety of settings without requiring a lot of retraining. By determining a numerical threshold score based on the anomaly's attributes, including frequency, intensity, and kind, this method is novel in that it can not only identify anomalies but also evaluate their seriousness. Because it enables security teams to rank answers based on the evaluated threat level, this severity score is essential for more efficient resource allocation and quicker reaction times.

Interface design for visual blind spots in cooperative driving

ABSTRACT Visual blind spots caused by occlusion from lead vehicles represent a significant latent risk factor contributing to traffic accidents. Although Level 2 (L2) autonomous driving systems can partially mitigate this issue, human drivers are still required to actively perceive potential hazards and understand the behaviour of the autonomous vehicle to ensure driving safety. Therefore, designing interactive interfaces that reduce the risks associated with visual blind spots is critical in autonomous driving scenarios. However, current research on this typical scenario remains relatively limited. This study proposes an innovative cooperative driving warning strategy, focussing on driving situations where the driver's line of sight is blocked by a lead vehicle. The strategy integrates environmental cues and behavioural information from the autonomous driving system through an augmented reality (AR) interface, aiming to facilitate efficient cooperation between human drivers and autonomous systems in perceiving visual blind spots. We systematically evaluated the proposed interface prototype using a driving simulator under L2 autonomous driving conditions. The results indicate that the interface significantly enhances drivers' situation awareness and reduces their reaction time to potential hazards. Additionally, the design improves the quality of human-machine cooperation by decreasing conflicts with the autonomous system, increasing trust in the system, and significantly boosting user satisfaction. This study provides new insights into the design of human-machine cooperative perception interfaces for blind spot scenarios and offers both theoretical foundations and practical implications for the future development of cooperative driving systems.

Frequency minimum inertia calculation of complex power systems based on an improved simulated annealing algorithm

In order to enhance the inertia situational awareness and stability of the power system in the context of the reduction of power system inertia and the challenge of frequency stability due to the high proportion of new energy access, the background and purpose of the study are stated in the introduction; methodologically, based on the frequency response model of the power system, we analyze the time and magnitude of the maximum frequency offset, clarify the factors affecting the frequency stability and construct the mathematical relationship, and then build a model for assessing the minimum inertia of the power system taking into account the virtual inertia of the new energy source. Then, based on the whole process of frequency response, we construct a power system minimum inertia assessment model taking into account the virtual inertia of new energy sources, and introduce an improved simulated annealing algorithm to solve the problem; the results validate the accuracy of the method through the IEEE-14 node model; the discussion section points out that this method provides a feasible solution for the inertia situational awareness of power system, which is helpful for the optimization of the operation, and also proposes that in the future we can take into account more uncertainties to improve the model and algorithm and to enhance the practicability and adaptability of this method. It is also suggested that more uncertainty factors can be considered in the future to improve the model and algorithm and enhance the practicality and adaptability.

Review on Information Fusion‐Based Data Mining for Improving Complex Anomaly Detection

ABSTRACTAnomaly predicated upon multiple distributed hybrid sensors frequently uses hybrid approaches, integrating techniques derived from statistical analysis, probability, data mining, machine learning, deep learning, and signal denoising. Many of these methods are based on the analysis of irregularities, data continuity, correlation, and data consistency, aiming to discern anomalous patterns from normal behavior. By leveraging these techniques information fusion aims to enhance situational awareness, detect potential threats or abnormalities, and improve decision‐making processes in complex environments. It addresses uncertainties by integrating data from diverse sources, thereby enhancing performance, and reducing dependency on individual sensors. This study examines applications based on single and multiple sensor data, revealing common strategies, identifying strengths and weaknesses, and potential solutions for detecting and diagnosing anomalies by analyzing low, large, and complex data derived from the context of homogeneous or heterogeneous systems. Information fusion techniques are evaluated for their performance on various levels of algorithm complexity. This in‐depth bibliographic study involved searching top indexing databases such as Web of Science and Scopus. The inclusion criteria were articles published between 2012 and 2024. The search capitalized on specific keywords as follows: “sensor malfunction,” “sensor anomaly,” “sensor failure,” “sensor fusion,” and “anomaly data mining.” Publications that did not strictly focus on analytical processing for anomaly detection, diagnosis, and prognosis in sensor data were excluded. In conclusion, the practice of information fusion promotes transparency by elucidating the process of combining information, thereby enabling the inclusion of multitude of perspectives, and aligning with established best practices in the field. Data deviation remains the primary criterion for detecting anomalies using mostly deep learning and extensively hybrid techniques. Nevertheless, state‐of‐the‐art algorithms based on neural networks still require further contextual interpretation and analysis. Functional safety and safety of intended functionality breaching can lead to decision‐making errors, physical harm, and erosion of trust in autonomous systems. This is due to the lack of interpretability in AI approaches, making it challenging to predict and understand the system's behavior under various conditions.

FIREGUIDE - An Advanced Emergency Response Platform Integrating Chatbot Assistance, Path Finder, and Alert Systems for Enhanced Fire Safety

Abstract - Fire emergencies in residential and urban settings pose significant risks, leading to potential injuries, fatalities, and property damage. While current safety measures, like fire alarms, provide basic alerts, they lack real-time, actionable guidance during crises. This work introduces FIREGUIDE, an innovative web- based platform designed to bridge these gaps. FIREGUIDE integrates a pre-trained chatbot for immediate assistance, a pathfinding tool for navigating safe evacuation routes, and a geolocation-based alert system to disseminate notifications within a 1-2 km radius. The chatbot module, powered by Natural Language Processing (NLP), offers contextually relevant guidance derived from historical fire incident data, empowering users to make sound decisions under stress. The alert system ensures prompt communication with nearby individuals, facilitating quick responses and coordinated evacuations. Meanwhile, the pathfinder tool provides dynamic, real-time route updates, enabling effective navigation to safety and reducing panic. By combining all these advanced tools, FIREGUIDE underscores the potential of modern technology to strengthen fire safety protocols and boost community resilience. This integrated approach aims to enhance situational awareness, minimize confusion, and improve response strategies during fire emergencies. Keywords — Fire Emergencies, Emergency and Real-Time Response Platform, Chatbot Assistance, Pathfinding Tool, Geolocation Alerting System, Community Resilience.

The impact of intraoperative non-technical skills training on scrub practitioners’ self-efficacy: a randomized controlled trial

BackgroundApproximately half of all adverse events occur in the operating room, highlighting the critical role of non-technical skills in operating rooms. Effective non-technical skills among operating room nurses can significantly reduce the occurrence of such events. Moreover, self-efficacy in non-technical skills may directly impact professional performance and patient safety. Therefore, this study aimed to investigate the impact of intraoperative non-technical skills training on scrub practitioners' self-efficacy.MethodsIn a randomized controlled trial, 30 scrub practitioners were assigned to the intervention group and 30 to the control group through random allocation. The intervention group underwent training in non-technical skills using a combined technique of lectures and simulated video scenarios delivered in two two-hour training sessions. Meanwhile, the control group received no training. The data collection tool was a two-part questionnaire. The first part collected demographic data (age, gender, work experience, and educational level), while the second part assessed scrub practitioners' self-efficacy in intraoperative non-technical skills. The questionnaire was administered online in two phases, with a one-month interval between them, through the Telegram application to the participants in both groups. The data were analyzed using descriptive statistics, independent t-tests, and paired t-tests.ResultsThe demographic variables of the intervention group did not show significant differences compared to the control group. The independent t-test revealed no significant difference in overall self-efficacy between the intervention and control groups before the training (P = 0.513). However, after the training, a statistically significant difference was observed (P = 0.025). There were no significant differences among the self-efficacy components between the intervention and control groups before the training (P > 0.05). However, after the training, self-efficacy in the two skills of situation awareness and communication and teamwork showed statistically significant differences (P < 0.05).ConclusionNon-technical skills are crucial for scrub practitioners to perform their tasks safely and efficiently. Training can enhance the self-efficacy of scrub practitioners in their non-technical skills. Therefore, it is necessary to incorporate non-technical skills training into the educational curriculum and continuing education programs for scrub practitioners.Trial registrationThe IRCT code (IRCT20150715023216N15) was obtained from the Iranian Clinical Trials Registry website on 2023/08/05 before sampling.

Facing Climate Change: How to Manage the Arising New Crises?

The increasing frequency and intensity of climate change-related disasters highlight the need to develop adaptive crisis management approaches. This paper explores the integration of R-IO Suite, a knowledge-based decision support system, with GAMA, a multi-agent simulation platform, to improve situational awareness and crisis response in a changing climate. This approach enables decision-makers and local communities to better understand, anticipate, and mitigate crisis impacts by combining real-time data modelling and predictive simulation. Although the proposed framework has been conceptually defined, its implementation and validation remain open key challenges. The French ANR-funded ATEsT project aims to overcome major technical obstacles— interoperability, semantic alignment, and temporal synchronisation—through concrete use cases. Future work will focus on implementing and testing this integration in realistic crisis scenarios, such as mega-fires and flash floods across different territories, to assess its effectiveness in improving decision-making processes and citizen engagement.

Risk Mitigation Measures Captured by a Tertiary Hospital's Disaster Simulation: An Observational Study.

To document the challenges experienced and adaptations made during a simulated hospital disaster, and consider the implications of the observations for hospital disaster preparedness. Nonparticipant observational assessment. Nonparticipant observations of an exercise simulating a disaster were undertaken by two researchers. The researchers shadowed triage team members, complementing this with observations of the Hospital-Emergency-Operations-Centre, theaters, wards, and Emergency Department subsections such as Resuscitation, Acute, Minor-Injuries-Clinic, Children's emergency, and Mental health. Field notes were coded line-by-line through an inductive thematic analysis, which synthesized both challenges and observed adaptations to those challenges. The major challenges observed were deaths due to lack of critical care equipment, management of high number of minor injuries, lack of situational awareness, shortage of orderlies, and difficulties in patient tracking and bed allocations. Observed adaptations included pediatricians' treatment of adult patients with minor injuries, fast-tracking triage through ranking, manual ventilation during transfers, and batching of patients requiring imaging to utilize limited orderlies for transfers. This observational study distills both challenges that clinicians may face in real disasters, and the improvisations that they can make to manage mass casualties. IMPLICATIONS FOR CLINICALPROFESSIONS: Research findings hold promising potential in enhancing clinicians' disaster preparedness by articulating specific interventions on mass-casualty management within limited resources. Unforeseen challenges arise when clinicians are confronted with disaster casualties. This study addresses that problem by not only preempting such challenges, but by also discussing practical solutions. The findings can enable a positive impact on clinicians' readiness for mass casualty influx. The 21-item checklist of the Standards for Reporting Qualitative Research (SRQR). Although this study was not focused on a patient population, our research institute incorporates healthcare consumers' advice in all our work.

Spline reconstruction of the informativeness template in correlation-extreme navigation tasks

The problem of correlation-extreme navigation is addressed based on the reconstructed spline template of informativeness, incorporating a priori information about the safe movement of vessels in conflict navigation spaces. The paper focuses on the practical implementation of intelligent ship motion control guided by the principle of analyzing the geophysical field geometry when enabling autonomous ship movement along an electronic spline trajectory. The study substantiates terrain-based navigation principles by comparing measured navigation parameters with a pre-created virtual informative template via an extreme correlation functional for continuous vessel location updates. An analysis of alternative positioning reveals the advantages and disadvantages of various autonomous map-aided navigation methods, emphasizing their potential accuracy. The hypothesis of spatial and surface field navigation as the sole alternative to satellite systems is examined. Alternative navigation is proposed as an assistive technology to complement traditional satellite positioning, ensuring maximum noise resistance and cybersecurity in operational situational awareness. Integrity monitoring is studied as a modern criterion for validating navigation information. The application of integrity is hypothesized to improve real-time iterative coordinate calculations in alternative positioning. A new procedure is developed to optimize grid approximation point distribution by determining the effective positions of sliding nodes. Computational implementation of a wide range of correlation-extreme navigation tasks is achieved through enhanced Pascal programs. The proposed algorithm, tested with spline function methods, provides harmonized assistance to navigators, extending situational awareness horizons for watch assistants navigating challenging scenarios.

Envisioning Cyber Situation Awareness Through Participatory Video Prototyping

Our digital societies are vulnerable to cyber crises. Without cyber-resilient organizations, vital societal functions may suffer incidents or loss of service. The diverse roles involved in cybersecurity decision-making require cyber situation awareness to uphold robust cybersecurity. Existing systems and processes supporting cyber situation awareness are not tailored to organizational needs, either at the role or the group level. This study explores the need for socio-technical system support, presenting common operational pictures supporting cyber situation awareness for staff handling cyberthreats. The participatory design method video prototyping was used to elicit needs from staff in a large, complex, public sector organization providing essential services. All participants have roles in cybersecurity crisis and incident management. Results from the video prototyping workshop suggest that cybersecurity staff need (i)~a single support system for incident management, and (ii)~a shared data repository underpinning (iii)~role-specific common operational pictures. The envisioned system support provides traceability and accountability.

Active shooters: The underutilisation of emergency operations centres - the need for a checklist to guide response and recovery.

The emergency operations centre (EOC) is a critical emergency response and recovery component that provides information management and resource allocation. EOCs are often used during all hazards; however, after reviewing over 25 after-action reports for active shooter incidents, they are frequently underutilised. Not activating or delaying activation can slow recovery efforts and lead to chaos for the first responders and the public due to a lack of situational awareness. Historical active shooter incidents, such as the San Bernardino attack, Uvalde school shooting and Aurora theatre shooting, highlight both the challenges and successes of EOC activations. Positive examples, including the Los Angeles International Airport (LAX) shooting and Pulse Nightclub attack, demonstrate how timely EOC activation improved resource coordination, victim services and public communications. A specialised active assailant checklist for EOC operations has remained largely absent even though the incidents pose a complex threat. The City of Murrieta and the City of Temecula worked to fill that void. They developed an 'Active Shooter EOC Checklist', informed by lessons learned from previous mass shootings and resources such as the 'United on Guns' protocol. The checklist guides the agencies through emergency operations, ensuring public communication, victim assistance, volunteer and donation management, recovery and other critical functions are not missed. This paper describes how EOC utilisation can streamline response operations, reduce fatalities and support community recovery efforts. This article is also included in The Business & Management Collection which can be accessed at https://hstalks.com/business/.

Enhancing Environmental Awareness of Deaf Individuals Using Deep Learning-Based Audio Classification

Abstract - Environmental sounds play a crucial role in human perception and situational awareness. However, individuals who are deaf or hard of hearing face significant challenges in detecting and interpreting such auditory cues, potentially compromising their safety and independence in everyday environments. This research presents a deep learning-based approach to real-time audio classification aimed at enhancing environmental awareness for deaf individuals. By leveraging convolutional neural networks (CNNs) and recurrent neural networks (RNNs), the system is trained to recognize a wide range of environmental sounds—such as sirens, alarms, doorbells, approaching vehicles, and human speech patterns. The classified audio cues are then translated into intuitive visual or haptic feedback through wearable or mobile devices. Experimental results demonstrate high accuracy in sound classification and low latency in alert delivery, making the system suitable for real-world deployment. This approach holds significant potential to bridge the sensory gap for the deaf community and foster greater independence and situational awareness. Keywords - CNN, RNN, Audio Classification, Deep Learning, Machine Learning

Advancing Early Wildfire Detection: Integration of Vision Language Models with Unmanned Aerial Vehicle Remote Sensing for Enhanced Situational Awareness

Early wildfire detection is critical for effective suppression efforts, necessitating rapid alerts and precise localization. While computer vision techniques offer reliable fire detection, they often lack contextual understanding. This paper addresses this limitation by utilizing Vision Language Models (VLMs) to generate structured scene descriptions from Unmanned Aerial Vehicle (UAV) imagery. UAV-based remote sensing provides diverse perspectives for potential wildfires, and state-of-the-art VLMs enable rapid and detailed scene characterization. We evaluated both cloud-based (OpenAI, Google DeepMind) and open-weight, locally deployed VLMs on a novel evaluation dataset specifically curated for understanding forest fire scenes. Our results demonstrate that relatively compact, fine-tuned VLMs can provide rich contextual information, including forest type, fire state, and fire type. Specifically, our best-performing model, ForestFireVLM-7B (fine-tuned from Qwen2-5-VL-7B), achieved a 76.6% average accuracy across all categories, surpassing the strongest closed-weight baseline (Gemini 2.0 Pro at 65.5%). Furthermore, zero-shot evaluation on the publicly available FIgLib dataset demonstrated state-of-the-art smoke detection accuracy using VLMs. Our findings highlight the potential of fine-tuned, open-weight VLMs for enhanced wildfire situational awareness via detailed scene interpretation.