Open-Insect: Benchmarking Open-Set Recognition of Novel Species in Biodiversity Monitoring
This addresses the challenge of monitoring highly diverse and poorly studied insect species for biodiversity conservation, though it is incremental as it focuses on benchmarking existing methods.
The paper tackles the problem of open-set recognition for detecting unknown insect species in biodiversity monitoring, introducing the Open-Insect dataset and benchmarking 38 algorithms, finding that simple post-hoc methods perform well as baselines.
Global biodiversity is declining at an unprecedented rate, yet little information is known about most species and how their populations are changing. Indeed, some 90% of Earth's species are estimated to be completely unknown. Machine learning has recently emerged as a promising tool to facilitate long-term, large-scale biodiversity monitoring, including algorithms for fine-grained classification of species from images. However, such algorithms typically are not designed to detect examples from categories unseen during training -- the problem of open-set recognition (OSR) -- limiting their applicability for highly diverse, poorly studied taxa such as insects. To address this gap, we introduce Open-Insect, a large-scale, fine-grained dataset to evaluate unknown species detection across different geographic regions with varying difficulty. We benchmark 38 OSR algorithms across three categories: post-hoc, training-time regularization, and training with auxiliary data, finding that simple post-hoc approaches remain a strong baseline. We also demonstrate how to leverage auxiliary data to improve species discovery in regions with limited data. Our results provide insights to guide the development of computer vision methods for biodiversity monitoring and species discovery.