Hybrid Approach for Effective Disaster Management Using Twitter Data and Image-Based Analysis

Modern technologies and social media offer exceptionally new approaches for fulfilling the needs of decision-makers during the disaster situation. The integration of real-time data with AI-based dynamic models generated from archive information acts as a powerful tool for decision-makers during the recovery operation of a disaster-affected area. Existing systems work either on images or tweets, but no system takes into account both the factors. The proposed novel approach demonstrates Image and Twitter-based flood mapping model which collates Images of flood zones and Twitter Data and presents them in a GIS-based system to facilitate an effective response to the disaster and provide subsequent relief operations. This system uses Named Entity Recognition to extract information from relevant tweets along with image segmentation and object detection to find out flood severity in a given flood site. This information is stored and displayed on a GIS interface to aid relief and recovery operations