Damage Identification in Social Media Posts using Multimodal Deep Learning: code and dataset

This project aims at predicting natural disasters using Machine Learning. This project was submitted as a part of Code.Fun.Do++ hackathon organised by Microsoft in 2018.

NLPrescue is an advanced Natural Language Processing system designed to detect and classify disaster-related tweets in real-time. Built with PyTorch and modern NLP techniques, it helps emergency responders quickly identify genuine disaster situations on social media platforms.

Alert Text Detector is an NLP-based model that detects alert messages from social media posts. It is built using BERTweet Base and trained on a dataset of 23,000 tweets (alert & non-alert). The model flags emergency-related messages and classifies tweets based on textual content.

Implementation of a Deep Neural Architecture to perform real-time semantic segmentation of forest fires in aerial imagery captured by drones.

Alert Text Detector is an NLP-based model that detects alert messages from social media posts. It is built using BERTweet Base and trained on a dataset of 23,000 tweets (alert & non-alert). The model flags emergency-related messages and classifies tweets based on textual content.