Search

Bangladesh is in the floodplains of the Ganges, Brahmaputra, and Meghna River delta, crisscrossed by an intricate web of rivers. Although the country is highly prone to flooding, the use of state-of-the-art deep learning models in predicting river water levels to aid flood forecasting is underexplored. Deep learning and attention-based models have shown high potential for accurately forecasting floods over space and time. The present study aims to develop a long short-term memory (LSTM) network and its attention-based architectures to predict flood water levels in the rivers of Bangladesh. The models developed in this study incorporated gauge-based water level data over 7 days for flood prediction at Dhaka and Sylhet stations. This study developed five models: artificial neural network (ANN), LSTM, spatial attention LSTM (SALSTM), temporal attention LSTM (TALSTM), and spatiotemporal attention LSTM (STALSTM). The multiple imputation by chained equations (MICE) method was applied to address missing data in the time series analysis. The results showed that the use of both spatial and temporal attention together increases the predictive performance of the LSTM model, which outperforms other attention-based LSTM models. The STALSTM-based flood forecasting system, developed in this study, could inform flood management plans to accurately predict floods in Bangladesh and elsewhere. [ABSTRACT FROM AUTHOR]

Earth observation and remote sensing necessitate the use of aerial imaging, and unmanned aerial vehicles (UAVs) are widely utilized for this purpose. Although UAVs can be costly, using inexpensive components could result in issues such as noise, defocusing, and motion blur. Utilizing advanced deep learning algorithms, these images can be effectively restored. However, this necessitates the use of physical effort since each image must be restored manually. Our work introduces a new pipeline that can automatically categorize the degradation types of aerial images and subsequently apply the necessary restoration techniques on degraded images. A lightweight adaptive degradation classifier model was proposed to handle input from images of varying sizes and provide guidance to a image restoration model for achieving the desired task. The degradation classifier is a Fully Convolutional Network capable of categorizing images of any size into 3 distinct classes. It offers guidance to the subsequent image restorer. The restoration model is a pre-trained transformer-based model that efficiently restores different types of image degradations. In addition, we have implemented a pipeline loss to assess the performance of our complete system. To build the system, a dataset is created by extracting cropped patches from high-resolution images obtained through Google Earth. Subsequently, the clean patches were artificially deteriorated by applying different filters to replicate the image degradations. The degradation classifier achieved a mean accuracy of 96.71% across 6 different test scenarios, with a standard deviation of 2.97. The mean pipeline loss was 24.95, with a standard deviation of 1.79. Furthermore, we showcased the efficacy of our pipeline in a practical implementation involving aerial imagery, specifically in the domain of vehicle detection. We conducted training on an object detection model and then evaluated the detection performance on both degraded and automatically restored images. Our experiments demonstrate the efficient performance of our method in automatically restoring degraded aerial images, even though the restoration models were not specifically trained on aerial images. In addition, we examined current limitations and difficulties in creating such systems and offered comprehensive perspectives on future advancements. • Inexpensive UAVs may suffer from various image quality degradations. • Large pre-trained image restorers can be utilized but require manual labor. • A novel lightweight model adaptively assesses images and guides the restorer. • A pipeline loss is constructed to evaluate the entire pipeline. • Real-world use cases are demonstrated and limitations are experimented. [ABSTRACT FROM AUTHOR]

Assessing flood risk is challenging due to complex interactions among flood susceptibility, hazard, exposure, and vulnerability parameters. This study presents a novel flood risk assessment framework by utilizing a hybridized deep neural network (DNN) and fuzzy analytic hierarchy process (AHP) models. Bangladesh was selected as a case study region, where limited studies examined flood risk at a national scale. The results exhibited that hybridized DNN and fuzzy AHP models can produce the most accurate flood risk map while comparing among 15 different models. About 20.45% of Bangladesh are at flood risk zones of moderate, high, and very high severity. The northeastern region, as well as areas adjacent to the Ganges-Brahmaputra-Meghna rivers, have high flood damage potential, where a significant number of people were affected during the 2020 flood event. The risk assessment framework developed in this study would help policymakers formulate a comprehensive flood risk management system. [ABSTRACT FROM AUTHOR]

The volume and availability of satellite image data has greatly increased over the past few years. But, during the transmission and acquisition of these digital images, noise becomes a prevailing term. When preprocessing the data for computer vision tasks, human experts often produce noise in the labels which can downturn the performance of learning algorithms drastically. This study is directed toward finding the effect of label noise in the performance of a semantic segmentation model, namely U-net. We collected satellite images of the Bangladesh marine region for four different time frames, created patches and segmented the sediment load into five different classes. The U-Net model trained with Dec-2019 dataset yielded the best performance and we tested this model under three types of label noise (NCAR - noise completely at random, NAR - noise at random and NNAR - noise not at random) while varying their intensity gradually from low to high. The performance of the model decreased slightly as the percentage of NCAR noise is increased. NAR is found to be defiant until 20° of rotation, and for NNAR, the model fails to classify pixels to its correct label for maximum cases. [ABSTRACT FROM AUTHOR]

Different epidemiological compartmental models have been presented to predict the transmission dynamics of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). In this study, we have proposed a fuzzy rule-based Susceptible-Exposed-Infectious-Recovered-Death (S E I R D) compartmental model considering a new dynamic transmission possibility variable as a function of time and three different fuzzy linguistic intervention variables to delineate the intervention and transmission heterogeneity on SARS-CoV-2 viral infection. We have analyzed the datasets of active cases and total death cases of China and Bangladesh. Using our model, we have predicted active cases and total death cases for China and Bangladesh. We further presented the correspondence of different intervention measures in relaxing the transmission possibility. The proposed model delineates the correspondence between the intervention measures as fuzzy subsets and the predicted active cases and total death cases. The prediction made by our system fitted the collected dataset very well while considering different fuzzy intervention measures. The integration of fuzzy logic in the classical compartmental model also produces more realistic results as it generates a dynamic transmission possibility variable. The proposed model could be used to control the transmission of SARS-CoV-2 as it deals with the intervention and transmission heterogeneity on SARS-CoV-2 transmission dynamics. [ABSTRACT FROM AUTHOR]

One of the challenges of training artificial intelligence models for classifying satellite images is the presence of label noise in the datasets that are sometimes crowd-source labeled and as a result, somewhat error prone. In our work, we have utilized three labeled satellite image datasets namely, SAT-6, SAT-4, and EuroSAT. The combined dataset consists of over 900,000 image patches that belong to a land cover class. We have applied some standard pixel-based feature extraction algorithms to extract features from the images and then trained those features with various machine learning algorithms. In our experiment, three types of artificial label noises are injected – Noise Completely at Random (NCAR), Noise at Random (NAR) and Noise Not at Random (NNAR) to the training datasets. The noisy data are used to train the algorithms, and the effect of noise on the algorithm performance are compared with noise-free test sets. From our study, the Random Forest and the Back-propagation Neural Network classifiers are found to be the least sensitive to label noises. As label noise is a common scenario in human-labeled image datasets, the current research initiative will help the development of noise robust classification methods for various relevant applications. [ABSTRACT FROM AUTHOR]

Globally, many studies on machine learning (ML)-based flood susceptibility modeling have been carried out in recent years. While majority of those models produce reasonably accurate flood predictions, the outcomes are subject to uncertainty since flood susceptibility models (FSMs) may produce varying spatial predictions. However, there have not been many attempts to address these uncertainties because identifying spatial agreement in flood projections is a complex process. This study presents a framework for reducing spatial disagreement among four standalone and hybridized ML-based FSMs: random forest (RF), k-nearest neighbor (KNN), multilayer perceptron (MLP), and hybridized genetic algorithm-gaussian radial basis function-support vector regression (GA-RBF-SVR). Besides, an optimized model was developed combining the outcomes of those four models. The southwest coastal region of Bangladesh was selected as the case area. A comparable percentage of flood potential area (approximately 60% of the total land areas) was produced by all ML-based models. Despite achieving high prediction accuracy, spatial discrepancy in the model outcomes was observed, with pixel-wise correlation coefficients across different models ranging from 0.62 to 0.91. The optimized model exhibited high prediction accuracy and improved spatial agreement by reducing the number of classification errors. The framework presented in this study might aid in the formulation of risk-based development plans and enhancement of current early warning systems. • Uncertainty exists in machine learning (ML)-based flood susceptibility models (FSMs). • Four standalone and hybridized ML-based FSMs were established. • An optimized model was developed to address uncertainties in ML-based FSMs. • Spatial association of flood probability between various models ranging from 0.62 to 0.91. • The optimized model reduced uncertainty and increased prediction accuracy. [ABSTRACT FROM AUTHOR]