Automatic Detection of Flood Severity Level from Flood Videos using Deep Learning Models

Abstract

Floods are one of the most destructive natural disasters, causing significant loss of life, property, and infrastructure. Accurate and timely assessment of flood severity levels is crucial for effective disaster management and mitigation. This project focuses on the automatic detection of flood severity levels from real-time flood videos using deep learning models. Leveraging advancements in computer vision, the proposed system extracts meaningful features from video frames to classify the severity of flooding into predefined levels. The framework integrates convolutional neural networks (CNNs) for feature extraction and recurrent neural networks (RNNs) for temporal analysis of video sequences. The model is trained on a dataset comprising diverse flood scenarios, ensuring robustness across varying environmental and geographical conditions. Preprocessing steps such as video frame extraction, resizing, and data augmentation enhance the model's accuracy. The proposed system is capable of providing rapid and reliable flood severity assessments, which can aid in real-time decision-making during flood emergencies. This project demonstrates the potential of deep learning in disaster management, offering a scalable and cost-effective approach to address one of the most pressing challenges in climate resilience. Future work includes integrating this system with IoT devices for continuous monitoring and expanding the dataset for improved generalization.