scholarly journals A Collaborative Region Detection and Grading Framework for Forest Fire Smoke Using Weakly Supervised Fine Segmentation and Lightweight Faster-RCNN

Forests ◽  
2021 ◽  
Vol 12 (6) ◽  
pp. 768
Author(s):  
Jin Pan ◽  
Xiaoming Ou ◽  
Liang Xu

Forest fires are serious disasters that affect countries all over the world. With the progress of image processing, numerous image-based surveillance systems for fires have been installed in forests. The rapid and accurate detection and grading of fire smoke can provide useful information, which helps humans to quickly control and reduce forest losses. Currently, convolutional neural networks (CNN) have yielded excellent performance in image recognition. Previous studies mostly paid attention to CNN-based image classification for fire detection. However, the research of CNN-based region detection and grading of fire is extremely scarce due to a challenging task which locates and segments fire regions using image-level annotations instead of inaccessible pixel-level labels. This paper presents a novel collaborative region detection and grading framework for fire smoke using a weakly supervised fine segmentation and a lightweight Faster R-CNN. The multi-task framework can simultaneously implement the early-stage alarm, region detection, classification, and grading of fire smoke. To provide an accurate segmentation on image-level, we propose the weakly supervised fine segmentation method, which consists of a segmentation network and a decision network. We aggregate image-level information, instead of expensive pixel-level labels, from all training images into the segmentation network, which simultaneously locates and segments fire smoke regions. To train the segmentation network using only image-level annotations, we propose a two-stage weakly supervised learning strategy, in which a novel weakly supervised loss is proposed to roughly detect the region of fire smoke, and a new region-refining segmentation algorithm is further used to accurately identify this region. The decision network incorporating a residual spatial attention module is utilized to predict the category of forest fire smoke. To reduce the complexity of the Faster R-CNN, we first introduced a knowledge distillation technique to compress the structure of this model. To grade forest fire smoke, we used a 3-input/1-output fuzzy system to evaluate the severity level. We evaluated the proposed approach using a developed fire smoke dataset, which included five different scenes varying by the fire smoke level. The proposed method exhibited competitive performance compared to state-of-the-art methods.

2019 ◽  
Vol 8 (4) ◽  
pp. 9126-9132

As we all know forests are the main source of oxygen and its protection is essential to sustain the human and animal race. Since we all learnt about the necessity of air, yet we lack at taking measures to protect our mother forest. Forest Fires are the main reason for the deforestation and destruction of trees and wildlife. Forest Fires are due to these two ways either by man-made or naturally caused. In either way we have to pay for the loss occurred because we have left with only certain area for the forest. So, we have to take measures to prevent forest fire at its early stage. The main aim of our project is to design and implement an IoT based hardware module that could detect the fire and prevent it by alerting the monitoring stations with an alert message and also provides location to the nearest base station. An automatic message will be sent to the nearest base station in addition to these, it has a 360 degrees rotation camera which helps to provide continuous surveillance. We can rotate the camera in any direction from the base station itself. A buzzer that alarms when the incident is happening and a water motor, this water motor will be on automatically. We can also find location where the incident is taking place with the help of Wi-Fi module. This device helps in identifying the fire at its early stage and helps in the prevention of spread all over the forest.


2018 ◽  
Vol 7 (3.12) ◽  
pp. 1312 ◽  
Author(s):  
Vinay Chowdary ◽  
Mukul Kumar Gupta ◽  
Rajesh Singh

Forest fire disasters have always been mankind’s constant and inconvenient companion since time immemorial. In the recent past years, managing crisis for example a large scale fire has become a very difficult and challenging task. Things that are common in most of the forest fire that occur at large scale are loss of life (human or animal), loss of vegetation, loss of flora and fauna, and communication failure (if any). Apart from causing a great loss to valuable natural resources of nature forest fire pose a greater risk not only to life of human being but also to the inhabitant’s such as wild life living in the forest. As per National Fire Danger Rating System (NFDRS), if a fire is detected within 6 minutes of its occurrence then it can be easily disposed-off before it turns into a large scale fire. For this a network that can detect fire at a very early stage is required. There are numerous techniques to detect the occurrence of forest fire and this article is dedicated towards reviewing detection techniques present in the literature. This work will give a bird’s eye view of the technologies used in automatic detection of forest fires and reviews almost all the detection techniques available in the literature. To the best of our knowledge this is the first time that almost all the techniques available in the literature are reviewed and considering almost all the parameters.


2003 ◽  
Vol 12 (2) ◽  
pp. 159 ◽  
Author(s):  
Andrei B. Utkin ◽  
Armando Fernandes ◽  
Fernando Simões ◽  
Alexander Lavrov ◽  
Rui Vilar

The feasibility and fundamentals of forest fire detection by smoke sensing with single-wavelength lidar are discussed with reference to results of 532-nm lidar measurements of smoke plumes from experimental forest fires in Portugal within the scope of the Gestosa 2001 project. The investigations included tracing smoke-plume evolution, estimating forest-fire alarm promptness, and smoke-plume location by azimuth rastering of the lidar optical axis. The possibility of locating a smoke plume whose source is out of line of sight and detection under extremely unfavourable visibility conditions was also demonstrated. The eye hazard problem is addressed and three possibilities of providing eye-safety conditions without loss of lidar sensitivity (namely, using a low energy-per-pulse and high repetition-rate laser, an expanded laser beam, or eye-safe radiation) are discussed.


2018 ◽  
Vol 10 (10) ◽  
pp. 102 ◽  
Author(s):  
Yi-Han Xu ◽  
Qiu-Ya Sun ◽  
Yu-Tong Xiao

Forest fires are a fatal threat to environmental degradation. Wireless sensor networks (WSNs) are regarded as a promising candidate for forest fire monitoring and detection since they enable real-time monitoring and early detection of fire threats in an efficient way. However, compared to conventional surveillance systems, WSNs operate under a set of unique resource constraints, including limitations with respect to transmission range, energy supply and computational capability. Considering that long transmission distance is inevitable in harsh geographical features such as woodland and shrubland, energy-efficient designs of WSNs are crucial for effective forest fire monitoring and detection systems. In this paper, we propose a novel framework that harnesses the benefits of WSNs for forest fire monitoring and detection. The framework employs random deployment, clustered hierarchy network architecture and environmentally aware protocols. The goal is to accurately detect a fire threat as early as possible while maintaining a reasonable energy consumption level. ns-2-based simulation validates that the proposed framework outperforms the conventional schemes in terms of detection delay and energy consumption.


Author(s):  
Zhentian Jiao ◽  
Youmin Zhang ◽  
Lingxia Mu ◽  
Jing Xin ◽  
Shangbin Jiao ◽  
...  

2018 ◽  
Author(s):  
Geraint Vaughan ◽  
Adam P. Draude ◽  
Hugo M. A. Ricketts ◽  
David M. Schultz ◽  
Mariana Adam ◽  
...  

Abstract. Layers of aerosol at heights between 2 and 11 km were observed with Raman lidars in the UK between 23 and 31 May 2016. A network of such lidars, supported by ceilometer observations, is used to map the extent of the aerosol and its optical properties. Spaceborne lidar profiles show that the aerosol originated from forest fires over Western Canada around 17 May, and indeed the aerosol properties – weak depolarisation and a lidar ratio at 355 nm in the range 35–65 sr – were consistent with long-range transport of forest fire smoke. The event was unusual in its persistence – the smoke plume was drawn into an atmospheric block that kept it above North-west Europe for nine days. Lidar observations show how the smoke layers became optically thinner during this period, but the lidar ratio and aerosol depolarisation showed little change.


2019 ◽  
Vol 65 (No. 4) ◽  
pp. 150-159
Author(s):  
Ding Xiong ◽  
Lu Yan

A smoke detection method is proposed in single-frame video sequence images for forest fire detection in large space and complex scenes. A new superpixel merging algorithm is further studied to improve the existing horizon detection algorithm. This method performs Simple Linear Iterative Clustering (SLIC) superpixel segmentation on the image, and the over-segmentation problem is solved with a new superpixel merging algorithm. The improved sky horizon line segmentation algorithm is used to eliminate the interference of clouds in the sky for smoke detection. According to the spectral features, the superpixel blocks are classified by support vector machine (SVM). The experimental results show that the superpixel merging algorithm is efficient and simple, and easy to program. The smoke detection technology based on image segmentation can eliminate the interference of noise such as clouds and fog on smoke detection. The accuracy of smoke detection is 77% in a forest scene, it can be used as an auxiliary means of monitoring forest fires. A new attempt is given for forest fire warning and automatic detection.


2018 ◽  
Vol 10 (12) ◽  
pp. 1992 ◽  
Author(s):  
Zixi Xie ◽  
Weiguo Song ◽  
Rui Ba ◽  
Xiaolian Li ◽  
Long Xia

Two of the main remote sensing data resources for forest fire detection have significant drawbacks: geostationary Earth Observation (EO) satellites have high temporal resolution but low spatial resolution, whereas Polar-orbiting systems have high spatial resolution but low temporal resolution. Therefore, the existing forest fire detection algorithms that are based on a single one of these two systems have only exploited temporal or spatial information independently. There are no approaches yet that have effectively merged spatial and temporal characteristics to detect forest fires. This paper fills this gap by presenting a spatiotemporal contextual model (STCM) that fully exploits geostationary data’s spatial and temporal dimensions based on the data from Himawari-8 Satellite. We used an improved robust fitting algorithm to model each pixel’s diurnal temperature cycles (DTC) in the middle and long infrared bands. For each pixel, a Kalman filter was used to blend the DTC to estimate the true background brightness temperature. Subsequently, we utilized the Otsu method to identify the fire after using an MVC (maximum value month composite of NDVI) threshold to test which areas have enough fuel to support such events. Finally, we used a continuous timeslot test to correct the fire detection results. The proposed algorithm was applied to four fire cases in East Asia and Australia in 2016. A comparison of detection results between MODIS Terra and Aqua active fire products (MOD14 and MYD14) demonstrated that the proposed algorithm from this paper effectively analyzed the spatiotemporal information contained in multi-temporal remotely sensed data. In addition, this new forest fire detection method can lead to higher detection accuracy than the traditional contextual and temporal algorithms. By developing algorithms that are based on AHI measurements to meet the requirement to detect forest fires promptly and accurately, this paper assists both emergency responders and the general public to mitigate the damage of forest fires.


2013 ◽  
Vol 2 (2) ◽  
pp. 171-177
Author(s):  
K. Nörthemann ◽  
J.-E. Bienge ◽  
J. Müller ◽  
W. Moritz

Abstract. Most huge forest fires start in partial combustion. In the beginning of a smouldering fire, emission of hydrogen in low concentration occurs. Therefore, hydrogen can be used to detect forest fires before open flames are visible and high temperatures are generated. We have developed a hydrogen sensor comprising of a metal/solid electrolyte/insulator/semiconductor (MEIS) structure which allows an economical production. Due to the low energy consumption, an autarkic working unit in the forest was established. In this contribution, first experiments are shown demonstrating the possibility to detect forest fires at a very early stage using the hydrogen sensor.


2021 ◽  
Vol I (I) ◽  
Author(s):  
Priyadharshini S

Forest fires are the most common threat in the woods. A combination of natural and human-made factors contributes to forest fires. Forest fires destroy trees, which are essential to produce oxygen, which we need to live. This new Zigbee-based wireless sensor network is being developed to overcome the limitations of existing technologies like the MODIS satellite-based detection system and a basic wireless sensor network. It's difficult to contain a forest fire that wasn't predicted or noticed in time. As a result, it's critical to catch a wildfire early enough before it spreads too far. Using a GSM device, the proposed method would gather data on forest conditions such as temperature, humidity, smoke, and flames, and deliver it to the appropriate authorities. There are three parts to the project's concept. Modules for sensors, gateways, and control centres make up the three sections. This project's main objective is to benefit others.


Sign in / Sign up

Export Citation Format

Share Document