Detecting and preventing forest fires is a significant issue for many nations. Various techniques have been suggested for keeping track of fire emergence.
A. S. Mahdi et al [1], 2022, demonstrate a survey of computer vision-based fire detection techniques. The recent machine learning models used, together with the datasets needed to build upcoming research projects for the wildfire detection system, are the main emphasis of this paper. Additionally, a comparison study highlights the advantages and disadvantages of the most critical aspects of the present approaches, such as detection accuracy. Alkhatib R et al [2], 2023, explores the application of various machine learning algorithms for forest fire detection and prediction. It discusses the advantages and limitations of algorithms such as decision trees, random forests, support vector machines, k-nearest neighbours, and neural networks. The study emphasises the importance of feature selection and data preprocessing to enhance model accuracy and efficiency. Abid F et al [3], 2021, review various forest fire detection techniques, including image processing, data mining, and machine learning approaches. It provides an overview of different datasets and the performance of algorithms applied. A. A. A Alkhatib [4], 2014, conducted a detailed review of Forest Fire Detection Techniques. Chowdary V and Gupta M K [5], 2018, a comprehensive survey of forest fire detection and monitoring techniques. It covers sensor-based approaches, remote sensing, wireless sensor networks, and machine-learning algorithms for early fire detection and monitoring. Grari, M et al [6], 2022, we undertook a comprehensive evaluation of the scholarly literature to detect and predict forest fires using IoT and machine/deep learning. The data analysis showed that temperature, humidity, CO, and light are the main factors in identifying and detecting forest fires. Additionally, we discovered that most communication channels used in this situation are based on one of these protocols: WiFi, ZigBee, or GSM.
I. Idrissi et al [7], 2022, developed a technique for Stratified IoT Deep Learning based Intrusion Detection System, which combines the concepts of multidisciplinary approaches of IoT and Deep Learning. M. Grari et al [8], 2022, discovered a methodology for early Wildfire Detection using Machine Learning Model Deployed in the Fog / Edge Layers of IoT. Rajan G V and Paul S [9], 2022, designed an algorithm for Forest Fire Detection Using Machine Learning. Sathish Kumar V. E et al [10], 2023, transfer learning is applied to pre-trained models like VGG16, InceptionV3, and Xception, which enables us to work with a smaller dataset and reduce computational complexity without compromising accuracy. With an accuracy rate of 98.72%, Xception outperformed all other models.
2.1 Existing System
The detection of a wildfire is primarily dependent upon 3 factors.
-
Oxygen Level: High oxygen content is required for any fire to take place. So, the higher the oxygen, the more is the probability of a wildfire taking place.
-
Temperature: For a fire to take place, heat is favourable. Hence, high temperature increases the probability of fire in any region.
-
Humidity: Humid weather is unfavorable for a fire, whereas a dry weather is. Therefore, the higher the humidity, the lower the probability of a fire.
2.2 Drawbacks of Existing System
Camera Surveillance: This approach uses drones or camera equipment to survey nearby forest cover for fires. However, the problem is, detection can only be done once a fire is actually started. It is also not economically feasible to cover large forest covers with cameras and drones.
Forest Fire Reservoirs: This creates water supplies near forest covers to extinguish fires early. This system again only works after a fire occurs and does not help detect the forest fire.
2.3 Proposed System
Machine learning models trained on data. So, we take real-life examples of forest fires that took place and collect the data before the fire took place, which is publicly available. We have the inputs as oxygen, humidity, temperature and the output as 0 or 1 based on whether a fire occurred. On creating a large enough dataset, we can create a trained machine learning model that can successfully predict the probability of a fire occurring in an area given the 3 parameters. Government can in that sense take necessary precautions for areas which high probability of a fire breaking out
2.4 Benefits of Proposed System
Once we get access to more data the machine learning model accuracy can be further increased
-
On a large scale this can be deployed by all forest authorities so that they have a prioritized list of places with places with maximum likelihood of a fire taking place at the top.
-
This can be combined with the web application to give a friendly interface for forest authorities, and this provides a way of smarter patrolling so that forests with greater likelihood of a fire taking place are given maximum patrolling and access to water supply
2.5 Scope
Forest or wildlife fires pose a significant threat to wildlife, and it is crucial to develop practical solutions to mitigate their impact. One of the main challenges in addressing forest fires is the early detection and prediction of these events, as once a fire starts, it can rapidly escalate and cause extensive, irreversible damage. Machine learning, a field focused on leveraging data to make predictions, can play a vital role. It is important to identify and model critical parameters that contribute to fire occurrence to develop a predictive model for forest fires. The government and appropriate agencies can then use this model to determine the best places to take preventative measures in the event of a fire. By implementing proactive measures based on these predictions, such as increased monitoring, enhanced fire prevention efforts, and resource allocation, the government can minimize the risk and impact of forest fires. This approach allows for targeted and efficient allocation of resources to areas at higher risk, reducing response time and improving overall fire management strategies.