edouard-rolland/YOLOv8-volcanic_plume-fuego_summit

Overview

This repository contains the trained weights and associated details for a YOLOv8s model fine-tuned to detect volcanic plumes and summit features of the Fuego Volcano in Guatemala. The dataset used for training can be found in the following Hugging Face dataset: volcanic-plume

Citation

@inproceedings{rolland2024volcanic,

    author    = {Edouard G. A. Rolland and Kasper A. R. Grøntved and Anders Lyhne Christensen and Matthew Watson and Tom Richardson},

    title     = { Autonomous {UAV} Volcanic Plume Sampling Based on Machine Vision and Path Planning},

    year      = { 2024 },

    note      = {Under review},

}

Acknowledgement

This work is supported by the WildDrone MSCA Doctoral Network funded by EU Horizon Europe under grant agreement no. 101071224, the Innovation Fund Denmark for the project DIREC (9142-00001B), and by the Engineering & Physical Sciences Research Council (UK) through the CASCADE (Complex Autonomous aircraft Systems Configuration, Analysis and Design Exploratory) programme grant (EP/R009953/1).

Object detection classes

['plume', 'summit']

Example of Predictions

The following video presents the model output for an entire flight. More details can be found in the paper above on the model training and performance.