The iWildCam 2018 Challenge Dataset

TL;DR

This paper introduces the iWildCam 2018 Challenge Dataset to evaluate the generalization of deep learning models for automatic annotation of camera trap images across different locations, aiming to scale biodiversity research.

Contribution

It provides a new dataset and benchmark to assess how well deep learning models trained on annotated images generalize to unseen locations.

Findings

Dataset enables evaluation of model generalization across locations

Benchmark results highlight challenges in cross-location annotation

Facilitates development of scalable biodiversity monitoring tools

Abstract

Camera traps are a valuable tool for studying biodiversity, but research using this data is limited by the speed of human annotation. With the vast amounts of data now available it is imperative that we develop automatic solutions for annotating camera trap data in order to allow this research to scale. A promising approach is based on deep networks trained on human-annotated images. We provide a challenge dataset to explore whether such solutions generalize to novel locations, since systems that are trained once and may be deployed to operate automatically in new locations would be most useful.