Vision Transformers versus convolutional neural networks for fine-grained orchid genus identification in a species-rich, data-poor flora: a controlled benchmark on the Orchidaceae of New Guinea
Abstract
New Guinea is the world's richest island flora (~2,856 orchid species), yet most species are represented by only a handful of photographs, far fewer than direct species-level classification requires. Methods for fine-grained identification in such species-rich, data-poor floras are needed, and it remains unclear which backbone architecture and pretraining strategy best support them. We built a two-stage system that first predicts the genus of a query photograph, then retrieves visually similar reference images of candidate species using FAISS. We compared four pretrained backbones -- two Vision Transformers (ViTs; DINOv2, BioCLIP 2) and two CNNs (ConvNeXt V2-L, EfficientNetV2-L) -- fine-tuned under an identical protocol on a fixed, species-stratified partition of 16,701 photographs spanning 120 genera and 1,350 species, assessing accuracy, calibration, error structure, species retrieval, and open-set detection of novel genera. DINOv2 attained the best genus performance (macro top-1 66.9%, 95% CI 63.7-70.6; global top-1 88.9%); both ViTs outranked both CNNs, and general-purpose self-supervised pretraining (DINOv2) outperformed domain-matched biological pretraining (BioCLIP 2) by 7.1 points of macro top-1. Errors concentrated on two abundant genera acting as error attractors. DINOv2 embeddings achieved species Recall@5 of 86.6% and genus Recall@5 of 98.7%; temperature scaling reduced every backbone's Expected Calibration Error to about 0.03; and a distance-based open-set gate flagged unseen genera (mean AUROC 0.958). A self-supervised Vision-Transformer backbone combined with embedding retrieval is an effective, deployable strategy for fine-grained identification in species-rich, data-poor floras. The system is released as an open web application (the New Guinea Orchid Identifier), offering a practical template for other hyperdiverse, under-documented taxa.
Related
- Vision Transformers Enable Advanced Plant Phenotyping in Controlled Environments
- Label-free identification of bacterial species using adapted vision transformers
- Genetic and floral variation in two colosely related Jumella spp. (Orchidaceae) in Mascarene Islands
- Sensitivity of Pulmonary 4D Flow MRI Biomarkers to Controlled Spatial and Temporal Degradation in a CFD Benchmark
- Three new species of Thelymitra (Diurideae, Orchidaceae) endemic to Aotearoa New Zealand.
BibTeX
@misc{saputra2026vision,
title = {{Vision Transformers versus convolutional neural networks for fine-grained orchid genus identification in a species-rich, data-poor flora: a controlled benchmark on the Orchidaceae of New Guinea}},
author = {Reza Saputra and Diah Harnoni Apriyanti and André Schuiteman and Kurt Metzger and Ashley Field and Katharina Nargar and William Edwards},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.24064},
eprint = {2609.24064},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.24064},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - UNPB
TI - Vision Transformers versus convolutional neural networks for fine-grained orchid genus identification in a species-rich, data-poor flora: a controlled benchmark on the Orchidaceae of New Guinea
AU - Reza Saputra
AU - Diah Harnoni Apriyanti
AU - André Schuiteman
AU - Kurt Metzger
AU - Ashley Field
AU - Katharina Nargar
AU - William Edwards
PY - 2026
DA - 2026/09/21
PB - arXiv
DO - 10.48550/arxiv.2609.24064
UR - https://arxiv.org/abs/2609.24064
AB - New Guinea is the world's richest island flora (~2,856 orchid species), yet most species are represented by only a handful of photographs, far fewer than direct species-level classification requires. Methods for fine-grained identification in such species-rich, data-poor floras are needed, and it remains unclear which backbone architecture and pretraining strategy best support them. We built a two-stage system that first predicts the genus of a query photograph, then retrieves visually similar reference images of candidate species using FAISS. We compared four pretrained backbones -- two Vision Transformers (ViTs; DINOv2, BioCLIP 2) and two CNNs (ConvNeXt V2-L, EfficientNetV2-L) -- fine-tuned under an identical protocol on a fixed, species-stratified partition of 16,701 photographs spanning 120 genera and 1,350 species, assessing accuracy, calibration, error structure, species retrieval, and open-set detection of novel genera. DINOv2 attained the best genus performance (macro top-1 66.9%, 95% CI 63.7-70.6; global top-1 88.9%); both ViTs outranked both CNNs, and general-purpose self-supervised pretraining (DINOv2) outperformed domain-matched biological pretraining (BioCLIP 2) by 7.1 points of macro top-1. Errors concentrated on two abundant genera acting as error attractors. DINOv2 embeddings achieved species Recall@5 of 86.6% and genus Recall@5 of 98.7%; temperature scaling reduced every backbone's Expected Calibration Error to about 0.03; and a distance-based open-set gate flagged unseen genera (mean AUROC 0.958). A self-supervised Vision-Transformer backbone combined with embedding retrieval is an effective, deployable strategy for fine-grained identification in species-rich, data-poor floras. The system is released as an open web application (the New Guinea Orchid Identifier), offering a practical template for other hyperdiverse, under-documented taxa.
ER -