UM-Deluge-Capstone / Annotated-Disaster-Imagery-Dataset

An imageset for object detection of annotated, street-level images from natural disasters.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Annotated Disaster Imagery Dataset

This repository serves as an accumulation of street level disaster imagery of buildings, houses, and architecture. One reoccurring theme in disaster response and recovery is the lack of imagery data on structural damage. Our research team collected and compiled datasets from various storm events such as hurricanes, tornados, and earthquakes. Altogether, the street level disaster dataset stands as a starting point for other researchers interested in continuing collection and annotation for machine learning research.

The dataset was annotated in Roboflow, an online platform for machine learning dataset creation and training. While this dataset was orginally designed for object detection, Roboflow can alter the format of this dataset for other algorithms.