# Detecting natural disasters, damage, and incidents in the wild

arXiv: https://arxiv.org/abs/2008.09188 · HTML: https://mlsanity.com/paper/2008.09188/

## Key facts

- Cited by 23 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 20 Aug 2020.

## Built on by (most cited first, 23 of 23)

- [RescueNet: A High Resolution UAV Semantic Segmentation Benchmark Dataset for Natural Disaster Damage Assessment](https://mlsanity.com/paper/2202.12361/) (2022)
- [Incidents1M: a large-scale dataset of images with natural disasters, damage, and incidents](https://mlsanity.com/paper/2201.04236/) (2022)
- [MEDIC: A Multi-Task Learning Dataset for Disaster Image Classification](https://mlsanity.com/paper/2108.12828/) (2021)
- [Harnessing Large Language Models for Disaster Management: A Survey](https://mlsanity.com/paper/2501.06932/) (2025)
- [Continual Learning for Smart City: A Survey](https://mlsanity.com/paper/2404.00983/) (2024)
- [Robust Training of Social Media Image Classification Models for Rapid Disaster Response](https://mlsanity.com/paper/2104.04184/) (2021)
- [NADBenchmarks -- a compilation of Benchmark Datasets for Machine Learning Tasks related to Natural Disasters](https://mlsanity.com/paper/2212.10735/) (2022)
- [RESPONSE: Benchmarking the Ability of Language Models to Undertake Commonsense Reasoning in Crisis Situation](https://mlsanity.com/paper/2503.11348/) (2025)
- [Deep-Disaster: Unsupervised Disaster Detection and Localization Using Visual Data](https://mlsanity.com/paper/2202.00050/) (2022)
- [VIDI: A Video Dataset of Incidents](https://mlsanity.com/paper/2205.13277/) (2022)
- [Building Damage Mapping with Self-PositiveUnlabeled Learning](https://mlsanity.com/paper/2111.02586/) (2021)
- [Do You See What I See? Capabilities and Limits of Automated Multimedia Content Analysis](https://mlsanity.com/paper/2201.11105/) (2021)
- [MS-DPPs: Multi-Source Determinantal Point Processes for Contextual Diversity Refinement of Composite Attributes in Text to Image Retrieval](https://mlsanity.com/paper/2507.06654/) (2025)
- [Landslide Detection in Real-Time Social Media Image Streams](https://mlsanity.com/paper/2110.04080/) (2021)
- [A Real-time System for Detecting Landslide Reports on Social Media using Artificial Intelligence](https://mlsanity.com/paper/2202.07475/) (2022)
- [CrashCar101: Procedural Generation for Damage Assessment](https://mlsanity.com/paper/2311.06536/) (2023)
- [Utilizing Post-Hurricane Satellite Imagery to Identify Flooding Damage with Convolutional Neural Networks](https://mlsanity.com/paper/2209.02124/) (2022)
- [Information Cascade Prediction under Public Emergencies: A Survey](https://mlsanity.com/paper/2404.01319/) (2024)
- [Social Media Data Mining of Human Behaviour during Bushfire Evacuation](https://mlsanity.com/paper/2512.01262/) (2025)
- [Architectural Insights for Post-Tornado Damage Recognition](https://mlsanity.com/paper/2602.14523/) (2026)
- [Survey on Disaster Management Datasets for Remote Sensing Based Emergency Applications](https://mlsanity.com/paper/2605.08196/) (2026)
- [Theia: Large-Scale Multimodal Captioning and Automated Validation of the Incidents1M Dataset for Data-Free Distillation](https://mlsanity.com/paper/2607.28269/) (2026)
- [RAPIDMap: Rapid Multi-Agent Pipeline for Interpretable Disaster Mapping from Satellite and Street-view Imagery](https://mlsanity.com/paper/2609.00046/) (2026)

## Abstract (arXiv)

Responding to natural disasters, such as earthquakes, floods, and wildfires, is a laborious task performed by on-the-ground emergency responders and analysts. Social media has emerged as a low-latency data source to quickly understand disaster situations. While most studies on social media are limited to text, images offer more information for understanding disaster and incident scenes. However, no large-scale image datasets for incident detection exists. In this work, we present the Incidents Dataset, which contains 446,684 images annotated by humans that cover 43 incidents across a variety of scenes. We employ a baseline classification model that mitigates false-positive errors and we perform image filtering experiments on millions of social media images from Flickr and Twitter. Through these experiments, we show how the Incidents Dataset can be used to detect images with incidents in the wild. Code, data, and models are available online at http://incidentsdataset.csail.mit.edu.