Aggregating User Input in Ecology Citizen Science Projects
Abstract
Camera traps (remote, automatic cameras) are revolutionizing large-scale studies in ecology. The Serengeti Lion Project has used camera traps to produce over 1.5 million pictures of animals in the Serengeti. To analyze these pictures, the Project created Snapshot Serengeti, a citizen science website where volunteers can help classify animals. To increase accuracy, each photo is shown to multiple users and a critical step is aggregating individual classifications. In this paper, we present a new aggregation algorithm which achieves an accuracy of 98.6%, better than many human experts. Our algorithm also requires fewer users per photo than existing methods. The algorithm is intuitive and designed so that nonexperts can understand the end results.
Cite
Text
Hines et al. "Aggregating User Input in Ecology Citizen Science Projects." AAAI Conference on Artificial Intelligence, 2015. doi:10.1609/AAAI.V29I2.19057Markdown
[Hines et al. "Aggregating User Input in Ecology Citizen Science Projects." AAAI Conference on Artificial Intelligence, 2015.](https://mlanthology.org/aaai/2015/hines2015aaai-aggregating/) doi:10.1609/AAAI.V29I2.19057BibTeX
@inproceedings{hines2015aaai-aggregating,
title = {{Aggregating User Input in Ecology Citizen Science Projects}},
author = {Hines, Greg and Swanson, Alexandra and Kosmala, Margaret and Lintott, Chris J.},
booktitle = {AAAI Conference on Artificial Intelligence},
year = {2015},
pages = {3975-3980},
doi = {10.1609/AAAI.V29I2.19057},
url = {https://mlanthology.org/aaai/2015/hines2015aaai-aggregating/}
}