Every year floods cause substantial threats to lives, livelihoods, agriculture and infrastructure. Rapid assessment of economic damage caused by floods is necessary for disaster management, resource allocation and policy making. In this study, we propose a novel method for calculating flood induced economic damage using before and after flood satellite imagery. By leveraging Vision Transformer techniques, we perform semantic segmentation to identify land cover changes after the disaster. By measuring the area loss per class and assigning economic value to each class, we provide a method to estimate the monetary damage due to the disaster. We used Segformer B3 for segmentation which is a model of the Vision Transformer and achieved much higher pixel accuracy$(0.98)$and$\text{mIoU}(0.53)$compared to state of the art segmentation models UNet and DeeplabV3+. Moreover, Segformer B3 demonstrated considerably higher computational efficiency compared to the two other models experimented. Our approach offers an innovative and automated solution for post disaster flood damage assessment.

Dhaka, a fast-growing metropolis with a population of approximately 21.7 million, is going through rapid social and economic advancement. However, rapid unplanned urbanization has resulted in substantial environmental damage and poses a severe public health risk. The Local Climate Zone (LCZ), a novel concept for urban heat island research and an-alyzing urban morphology, provides a standardized categorization system for the urban environment. This study aims to investigate the LCZ mapping of two densely populated Asian cities: Dhaka, Bangladesh, and Kolkata, India, using two machine learning clustering algorithms, K-means and Iterative Self-Organizing Data Analysis Technique Algorithm (ISODATA). The research primarily focuses on Dhaka, but Kolkata is chosen as a comparative city to deliver a more in-depth analysis because the two cities share similar geographical and environmental characteristics. The study reveals that ISODATA is particularly good at classifying specific urban types, especially distinguishing between open and compact mid-rise.

Bangladesh is the fourth largest rice-producing country in the world. Agriculture plays a vital role in the country's economy. One of the major obstacles in rice production is rice paddy diseases. In this paper, we develop a deep learning-based system to detect rice paddy diseases. In the first step, a rice paddy image dataset is analyzed and preprocessed for classification. To build the classifier, we use the Efficient Net B3 Convolution Neural Network (CNN) model. Next, we train a new model using segmented rice paddy disease-affected areas to detect affected regions using MASK Recurrent Convolutional Neural Network (Mask RCNN). For the classification methods, we obtain an accuracy of nearly ~99%. For segmentation, the loss value of the class, bounding box, and mask are 0.09, 0.29, 0.30. The mean Average Precision(mAP) of the segmentation is around ~89%.

In warehouses, specialized agents need to navigate, avoid obstacles and maximize the use of space in the warehouse environment. Due to the unpredictability of these environments, reinforcement learning approaches can be applied to complete these tasks. In this paper, we propose using Deep Reinforcement Learning (DRL) to address the robot navigation and obstacle avoidance problem and traditional Q-learning with minor variations to maximize the use of space for product placement. We first investigate the problem for the single robot case. Next, based on the single robot model, we extend our system to the multi-robot case. We use a strategic variation of Q-tables to perform multi-agent Q-learning. We successfully test the performance of our model in a 2D simulation environment for both the single and multi-robot cases.

This paper presents a computer vision-based system for traffic offense detection. The system detects traffic offenses such as speed limit violations, unauthorized vehicles, traffic signal violations, unauthorized parking, wrong-way driving, and motorbike riders without helmets. The traffic offense detection system consists of a pipeline of four different modules. These are a vehicle detection module, a vehicle classification module, a vehicle tracking module, and a traffic offense detection module. Vehicles on the roads are detected in the vehicle detection module using visual data such as live camera feed. Next, after the vehicles are detected, they are classified into different classes using a vehicle classification module. A vehicle tracking module is developed to track the vehicle as it moves through the traffic. Lastly, we have implemented a traffic offense detection module that analyzes traffic patterns and detects different types of traffic violations in real-time. The entire system is implemented using OpenCV Deep Neural Network (DNN) module. We have used YOLOv4 to detect vehicles on the roads with high accuracy. For motorbike riders without helmets, we have used a fast YOLOv4-tiny model. The DeepSORT algorithm is used to track vehicles in real-time. Obtained accuracies are 86% in YOLOv4 for vehicle detection and 92% in YOLOv4-tiny for helmet detection.

Continuous launching of new satellites and increasing numbers of space missions is making space a congested environment. Collision with space debris or other satellites is now a real problem for satellites with the problem more acute in highly trafficked orbits. Thus, mission operators and space agencies are in need of high accuracy collision avoidance systems for spacecrafts and satellites. This paper focuses on tackling the satellite collision problem by implementing a collision avoidance system using neural networks and relevant machine learning techniques. The primary model is based on General Regression Neural Network (GRNN) and the secondary models are based on Artificial Neural Networks (ANN), Random Forest Regression & Support Vector Regression techniques. The dataset used in this paper is collected from the ESA (European Space Agency) which contains the events of risk assessment or in other words, Conjunction Data Messages (CDM). The proposed collision avoidance system predicts the collision risk percentage between target (a satellite of interest) and chaser (space debris or another satellite) objects. The predicted risk enables the target to maneuver accordingly and ultimately avoid collision with the chaser object. The GRNN algorithm uses lazy learning which does not require iterative training and makes predictions based on stored parameters. The training data has been normalized before applying the algorithm as GRNN network is sensitive to high deviation among input features. However, the GRNN model predicts the risk of collision between the target & the chaser object with an MSE (Mean Squared Error) of 11% which means the model predicts the risk of collision with 89% accuracy and this 89% risk can give enough confidence factor to the concerned authority to take necessary evasive maneuvers. This is reliable enough and lower than other models’ MSE to consolidate the fact that the GRNN model is best fit for our dataset.

In this study, we use the combination of remote sensing and deep learning techniques to predict agricultural yield of potato one of the most consumed items all over the world. Time-Series analysis of rainfall using an LSTM (Long Short Term Memory) network is also exhibited in this study to help better understand the changes in climate and crop yield. The study was carried out in Munshiganj district of Bangladesh, which is the main contributor of potatoes in the country. With our results, we exhibit how the changes in seasonal rainfall affects the country's economy as well as livelihood since majority of the people's livelihood depends on agriculture.