A modified YOLOv5 architecture for efficient fire detection in smart cities

Fire hazards pose a significant risk to human lives, infrastructure, and the environment, necessitating the development of efficient fire detection and localization systems. Traditional methods, including smoke and heat sensors, suffer from high false alarm rates and delayed response times. In recent years, deep learning has emerged as a transformative approach, leveraging computer vision to enhance accuracy in fire detection and localization. This paper provides a comprehensive survey of deep learning techniques employed for fire detection, including Convolutional Neural Networks (CNNs), object detection models such as YOLO and Faster R-CNN, and hybrid approaches integrating multimodal data. The study further explores publicly available fire datasets, preprocessing techniques, and performance evaluation metrics. Additionally, key challenges such as real-time processing constraints, environmental variability, and model generalization are discussed. The paper concludes with an outlook on future advancements, including lightweight AI models, multi-sensor fusion, and synthetic dataset generation for robust fire detection systems.

In the current context of climate change and demographic expansion, one of the phenomena that humanity faces are the suburban wildfires. To prevent the occurrence of suburban forest fires, fire risk assessment and early fire detection approaches need to be applied. Forest fire risk mapping depends on various factors and contributes to the identification and monitoring of vulnerable zones where risk factors are most severe. Therefore, watchtowers, sensors, and base stations of autonomous unmanned aerial vehicles need to be placed carefully in order to ensure adequate visibility or battery autonomy. In this study, fire risk assessment of an urban forest was performed and the recently introduced 360-degree data were used for early fire detection. Furthermore, a single-step approach that integrates a multiscale vision transformer was introduced for accurate fire detection. The study area includes the suburban pine forest of Thessaloniki city (Greece) named Seich Sou, which is prone to wildfires. For the evaluation of the performance of the proposed workflow, real and synthetic 360-degree images were used. Experimental results demonstrate the great potential of the proposed system, which achieved an F-score for real fire event detection rate equal to 91.6%. This indicates that the proposed method could significantly contribute to the monitoring, protection, and early fire detection of the suburban forest of Thessaloniki.

In an emergency, such as fire in a building, visually impaired people are prone to danger more than non-impaired people, for they cannot be aware of it quickly. Current fire detection methods such as smoke detector is very slow and unreliable. But by using vision sensor instead, fire can be proven to be detected much faster as shown in our experiments. Previous studies have applied various image processing and machine learning techniques to detect fire, but they usually don’t generalize well because those techniques use hand-crafted features. With the recent advancements in the field of deep learning, this research can be conducted to help solve the problem by using deep learning-based object detector to detect fire. Such approach can learn features automatically, so they can usually generalize well to various scenes. We introduced two object detection models (R1 and R2) with slightly different model’s complexity. R1 can detect fire at 90% average precision and 85% recall at 33 FPS, while R2 has 90% average precision and 61% recall at 50 FPS. The reason why we introduced two models is because we want to have a benchmark comparison as no other research on fire detection with similar techniques exists. We also want to give two model choices when we wish to integrate the model into an IoT platform.

<strong class="journal-contentHeaderColor">Abstract.</strong> With the advancement of computer vision, artificial intelligence, and remote sensing technologies, deep learning algorithms are increasingly used in terrestrial, airborne, and spaceborne-based fire detection systems. The performance and generalization of these data-driven fire detection algorithms, however, are restricted by the limited number and source of fire detection datasets. A large-scale fire detection benchmark dataset covering complex and varied fire scenarios is urgently needed. This work constructs a 100,000-level Flame and Smoke Detection Dataset (FASDD) based on multi-source heterogeneous flame and smoke images. It holds rich variations in image size, resolution, illumination (day and night), scenario (indoor and outdoor), image range (far and near), viewing angle (top view and side view), platform (surveillance cameras, drones, and satellites), and data source (Internet, social media, and open-access fire datasets). To the best of our knowledge, FASDD is currently the most versatile and comprehensive dataset for fire detection. It provides a challenging benchmark to drive the continuous evolution of fire detection models. Additionally, we formulate a unified workflow for preprocessing, annotation, and quality control of fire samples. Out-of-the-box annotations are published in four different formats for training deep learning models. Extensive performance evaluations based on classical methods show that most of the object detection models trained on FASDD can achieve satisfactory fire detection results, and especially YOLOv5x achieves nearly 80 % mAP@0.5 accuracy on heterogeneous images spanning two domains of computer vision and remote sensing. And the application in wildfire location demonstrates that deep learning models trained on our dataset can be used in recognizing and monitoring forest fires. Deep learning models trained with FASDD can be simultaneously deployed on satellites, drones, and ground sensors, thus realizing collaborative fire observation and detection in a space-air-ground integrated network environment. The dataset is available from the Science Data Bank website at <a href="https://doi.org/10.57760/sciencedb.j00104.00103" target="_blank" rel="noopener">https://doi.org/10.57760/sciencedb.j00104.00103</a> (Wang et al., 2022a).

Facile construction of a flexible smart core-sheath flax yarns with temperature-responsive resistance for ultra-fast fire-alarm response

A fire disaster that does not save can certainly cause losses, both in the form of objects and casualties. This occurs for several reasons: late information obtained from the fire department or the owner's ignorance at the time of a fire. In this study, a fire early detection system was built using smoke, heat, and gas sensors based on an SMS gateway and an alarm. This system is used to provide information about fire detection as early as possible to protect against fire disasters. With this system, the potential and risk of fire can be reduced. This system is used to identify potential fires that occur in housing. Several experiments were carried out with fire simulations to get the reaction from the sensors used. Covers smoke testing, temperature testing, gas testing, and SMS message responses from various providers. This research produces a fire early warning system that provides SMS and alarm alerts.

Fast fire flame detection in surveillance video using logistic regression and temporal smoothing

In practical scenarios such as industrial plants, warehouses, laboratories, residential complexes, and forest monitoring zones, there is a critical requirement for intelligent systems capable of early fire detection, real-time monitoring, and rapid response without risking human lives. These environments demand autonomous or remotely controlled solutions that can operate efficiently under extreme conditions. Traditional firefighting methods rely heavily on manual intervention, which can be dangerous, time-consuming, and often delayed due to accessibility challenges. Furthermore, conventional systems lack real-time monitoring, automated detection, and precise fire targeting mechanisms, reducing their effectiveness in critical situations. To address these challenges, the proposed Fire Bot Sentinel utilizes the ESP32 microcontroller integrated with robotic and IoT technologies to develop an intelligent firefighting system. The system employs flame sensors to detect fire presence and automatically activates a water pump to extinguish flames. A mobile robotic platform enables the system to navigate toward the fire source, while a servo motor controls the direction of water spraying for precise targeting. The integration of an ESP32-CAM allows real-time video streaming and remote monitoring of fire conditions. An LCD display provides system status updates, and the battery-powered design ensures portability in hazardous environments. This intelligent system enhances early fire detection, minimizes human risk, enables rapid response, and provides a reliable and scalable solution for modern fire safety and disaster management applications.

During the last decade, infrared (IR) sensors have proved to be a useful tool in forest fire fighting. The first applications came from satellite sensors originally devised for meteorological applications, like the Advanced Very High Resolution Radiometer (AVHRR) onboard the NOAA satellites. Now, a new generation of specific fire satellites is planned that promises early fire outbreak detection and monitoring. The success of satellite IR sensors has encouraged many forest services to use ground-based IR sensors for fire detection, typical devices being IR cameras placed on rotary platforms on lookout towers. These ground-based applications have been confined nearly always to fire detection, but IR sensors can provide much more information on fires. This information may be very useful to get a better understanding of fire properties and behaviour, in particular by providing inputs and experimental validation for the theoretical models of fire spread. This paper focuses on the problem of how to get quantitative information on relevant physical parameters of forest fires by using IR cameras. Traditionally, physical information on forest fires is obtained laboratory fires with thermocouples and by measuring the fuel mass loss. This is a very labour-demanding task, that requires to deploy

Early fire detection is an important factor that can reduce economic and environmental damage and reduce the number of victims. Despite the growing popularity of neural networks as a modern method for solving problems in computer vision, methodological problems often arise in works in this subject area, leading to a decrease or complete devaluation of practical results. This study is devoted to finding such problems among existing works on fire detection. The first section contains a contrast analysis of two works, during which 11 meta-criteria were identified to assess the quality of studies. The second section contains an overview of several works devoted to fire detection in various conditions, both by «classical» methods and using convolutional neural networks. The importance of the correct choice of metrics, the need to choose a model as a process, and a full description of the source data are shown.

I. ABSTRACT Fire hazards pose a significant threat to life and property, making early detection crucial for minimizing damage. This project focuses on AI-based fire detection and alarming using computer vision and deep learning techniques. The system employs CCTV cameras or laptop cameras to capture real-time video feeds, which are analysed using Python, OpenCV, and deep learning models to detect fire accurately. A Convolutional Neural Network (CNN) or YOLO (You Only Look Once) object detection model is trained on fire datasets to distinguish fire from non-fire scenarios, reducing false alarms. Once fire is detected, the system triggers an alarm sound and can also send notifications via SMS, email, or IoT-based alerts to concerned authorities. This AI-powered solution is designed for high accuracy and efficiency, making it suitable for smart surveillance, industrial safety, and residential fire monitoring. The integration of edge computing devices like Raspberry Pi or Jetson Nano can further enhance real-time processing. The proposed system aims to improve fire safety measures by providing automated, real-time fire detection, reducing response time, and preventing large-scale disasters. II. KEYWORDS Fire Detection, Real-time Fire Monitoring, Computer Vision (CV).

Forest fire (wildfire) is one of the common hazards that is accrued in the forest. Fire monitoring has three phases: pre-fire (take appropriate action for fire control), during fire (detection of fire and planning to control fire), post-fire (damage assessment and mitigation planning). In older days, manually fire detection approach is used. In current days, satellite-based surveillance system is used to detect forest fire but this works when fire is spread in the large area. So these techniques are not efficient. According to a survey, approximately 80% losses are accrued in the forest due to the late detection of fire. So to overcome this problem, we use the Internet of things technology. In this paper, early fire detection model has been proposed with the help of the Raspberry Pi microcontroller and required sensors. Centralized server is used for storing the data and analyzing that data. Feed-forward fully connected neural network is used for prediction purpose. Then, an alert message is sent to the admin and to the people within the proximity.

Research on multi-detector real-time fire alarm technology based on signal similarity

Aiming at the high false alarm rate when using single sensor to detect fire in aircraft cabin, a multisensor data fusion method is proposed to detect fire. First, the weights of multiple factors, that is, temperature, smoke concentration, CO concentration, and infrared ray intensity in the event of fire, were calculated by using the improved analytic hierarchy process (AHP) method on each sensor node of wireless sensor network, and the probability of fire event in the cabin was evaluated by multivariable‐weighted fusion method. Second, based on the mutual support among the evaluation data of fire probabilities of each node, the adaptive weight coefficient is assigned to each evaluation value, and the weighted fusion of all evaluation values of each node is conducted to obtain the fire probability. In the end, compared to the threshold of probability, the fire alarm is determined. Comparing the proposed algorithm to the grey fuzzy neural network fusion algorithm and fuzzy logic fusion algorithm in terms of the time consumption for fire detection and sending alarm and the accuracy of fire alarm perspectives, the experiments demonstrate that the proposed fire detection algorithm can detect the fire within 10s and reduce the false alarm rate to less than 0.5%, which verifies the superiority of the algorithm in promptness and accuracy. In the meanwhile, the fault tolerance of the algorithm is proved as well.

Evaluation of image fire detection algorithms based on image complexity