A training protocol for human classification of Asian elephant images from trail cameras
Trail cameras have become ubiquitous tools for ecological data collection over recent decades. Despite progress in the development of automated algorithms and artificial intelligence for image classification, our ability to process large volumes of data remain limited by the need for trained human observers to make refined judgements. We provide guidance on placement of trail cameras for…
Trail cameras have become essential tools for collecting ecological data over the past few decades. Although automated algorithms and artificial intelligence have advanced, the sheer volume of data still necessitates human intervention for refining judgments. This article outlines a protocol for training and testing human observers to classify images of Asian elephants (Elephas maximus) based on age, sex, and group composition.
The protocol was developed using 14,007 images collected from six trail cameras in Udawalawe National Park, Sri Lanka, between 2017 and 2019. Initially, three expert observers trained a group of four inexperienced participants. They engaged in an iterative process to create a comprehensive protocol document.
The protocol was subsequently tested on a separate group of six subjects, each of whom classified 350 test images in four sequential batches. The accuracy and precision of these test subjects were measured using quantitative metrics. Compared to the expert observers, the test subjects achieved a fair level of precision, with a Fleiss kappa score of 0.247, and an 82.6% accuracy rate.
This approach can be adapted to classify images of other species and contexts, providing a high-throughput workflow for extracting useful data from large volumes of trail camera images.
Written by urgent.news from bioRxiv's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
