Search

Wildfire early detection and prevention had become a priority. Detection using Internet of Things (IoT) sensors, however, is expensive in practical situations. The majority of present wildfire detection research focuses on segmentation and detection. The developed machine learning models deploy appropriate image processing techniques to enhance the detection outputs. As a result, the time necessary for data processing is drastically reduced, as the time required rises exponentially with the size of the captured pictures. In a real-time fire emergency, it is critical to notice the fire pixels and warn the firemen as soon as possible to handle the problem more quickly. The present study addresses the challenge mentioned above by implementing an on-site detection system that detects fire pixels in real-time in the given scenario. The proposed approach is accomplished using Deeplabv3+, a deep learning architecture that is an enhanced version of an existing model. However, present work fine-tuned the Deeplabv3 model through various experimental trials that have resulted in improved performance. Two public aerial datasets, the Corsican dataset and FLAME, and one private dataset, Firefront Gestosa, were used for experimental trials in this work with different backbones. To conclude, the selected model trained with ResNet-50 and Dice loss attains a global accuracy of 98.70%, a mean accuracy of 89.54%, a mean IoU 86.38%, a weighted IoU of 97.51%, and a mean BF score of 93.86%. [ABSTRACT FROM AUTHOR]

• The current study extracts higher-order features from fire images. • A new set of data is correctly labelled for classifying fire and non-fire. • Information-theoretic feature selection is adopted to minimize computational cost. • The SVM performs classification with an RBF kernel. • The model draws an overall accuracy of 96,21%, and a specificity of 97,99%. • The model draws an f-measure and g-mean values of 96,13% and 96,19% respectively. In recent years, wildfires and forest fires have ravaged millions of hectares of the forest all over the world. Recent technological breakthroughs have increased interest in computer vision-based fire classification that classifies fire and non-fire pixels from image or video datasets. Fire pixels from an image or video can be classified using either a traditional machine learning approach or a deep learning approach. Presently, the deep learning approach is the mainstream in forest fire detection studies. Although deep learning algorithms can handle vast amounts of data, they ignore the variation in complexity among training samples and as a result, their training model performance is limited. Furthermore, deep learning approaches with little data and features perform poorly in real-world challenging fire scenarios. As a result, the current study adopts a machine learning technique to extract higher-order features from the processed images from the publicly available datasets: Corsican dataset and FLAME, and a private dataset: Firefront_Gestosa, for classifying fire and non-fire pixels. It should be emphasized that in machine learning applications, handling multidimensional data to train a model is challenging. Feature selection is used to overcome this problem by removing redundant or irrelevant data that has an impact on the model's performance. In this paper, information-theoretic feature selection approaches are used to choose the most important features for classification while minimizing the computational cost. The traditional machine classifier, Support Vector Machine (SVM) is adopted in the present work, that works on the discriminative features input selected from the feature selection technique. The SVM performs the classification of fire and non-fire pixels with a Radial Basis Function (RBF) kernel, and the model's performance is measured using assessment measures such as overall accuracy, sensitivity, specificity, precision, recall, F-measure, and G-mean. The model draws an overall accuracy of 96,21%, a sensitivity of 94,42%, a specificity of 97,99%, a precision of 97,91%, a recall of 94,42%, an f-measure and g-mean values of 96,13% and 96,19% respectively. [ABSTRACT FROM AUTHOR]

Device-free human gesture recognition (HGR) using commercial off the shelf (COTS) Wi-Fi devices has gained attention with recent advances in wireless technology. HGR recognizes the human activity performed, by capturing the reflections of Wi-Fi signals from moving humans and storing them as raw channel state information (CSI) traces. Existing work on HGR applies noise reduction and transformation to pre-process the raw CSI traces. However, these methods fail to capture the non-Gaussian information in the raw CSI data due to its limitation to deal with linear signal representation alone. The proposed higher order statistics-based recognition (HOS-Re) model extracts higher order statistical (HOS) features from raw CSI traces and selects a robust feature subset for the recognition task. HOS-Re addresses the limitations in the existing methods, by extracting third order cumulant features that maximizes the recognition accuracy. Subsequently, feature selection methods derived from information theory construct a robust and highly informative feature subset, fed as input to the multilevel support vector machine (SVM) classifier in order to measure the performance. The proposed methodology is validated using a public database SignFi, consisting of 276 gestures with 8280 gesture instances, out of which 5520 are from the laboratory and 2760 from the home environment using a 10 × 5 cross-validation. HOS-Re achieved an average recognition accuracy of 97.84%, 98.26% and 96.34% for the lab, home and lab + home environment respectively. The average recognition accuracy for 150 sign gestures with 7500 instances, collected from five different users was 96.23% in the laboratory environment. [ABSTRACT FROM AUTHOR]