Most of the time, natural disasters tend to interfere with the availability of communication infrastructure leading to a delay in coordinating the relief efforts. Therefore, ReliefLink is a mobile application designed to coordinate the activities of the disaster victims, volunteer agencies, non-governmental orga nizations and administrators within one platform. The mobile application will comprise MobileNetV2 for analyzing images uploaded by the survivors through their smartphone while XLM RoBERTa based natural language processing models will be used in classifying texts and verbal reports. Built with Flutter and Firebase tools, ReliefLink utilizes the offline capability of Firebase Cloud Firestore to temporarily use the data contained in the core application. The data is synchronized with the cloud as soon as internet becomes available making sure that the accuracy of the data in the cloud storage is maintained. Moreover, the use of location services and geoclustering makes it easier for volunteers to determine the priority areas. Role-based dashboards ensure that communication, reporting, volunteer coordination and relief management are carried out effectively among the users of the application. Unlike the existing disaster management applica tions which are limited to reporting and coordination purposes respectively, the proposed mobile application integrates different technologies including artificial intelligence, offline data support, cloud synchronization, location aware decisions, and role-based collaboration into one mobile application.

Disaster management, artificial intelligence, MobileNetV2, XLM-RoBERTa, natural language processing, image classi fication, Flutter, Firebase Cloud Firestore, OpenStreetMap, offline data support