Self-Supervised Learning of Plant Image Representations

2026-04-30

AI Summary

This work addresses the limitations of existing plant identification methods, which heavily rely on expert annotations, and the suboptimal performance of generic self-supervised learning approaches on fine-grained plant image tasks. To overcome these challenges, the authors propose a domain-adapted self-supervised representation learning framework tailored for plant imagery. They introduce targeted data augmentation strategies—including affine transformations and color tone separation—and integrate them into a SimDINOv2 framework built upon Vision Transformer (ViT) architectures, trained on the Plantae subset of iNaturalist 2021. Experimental results demonstrate that the proposed method matches or exceeds strong supervised baselines such as Pl@ntCLEF and BioCLIP in few-shot downstream tasks, offering a scalable and annotation-efficient solution for biodiversity monitoring.

Abstract

Automated plant recognition plays a crucial role in biodiversity monitoring and conservation, yet current approaches rely heavily on supervised learning, which is limited by the availability of expert-labeled data. Self-supervised learning (SSL) offers a scalable alternative, but existing methods and training protocols are largely designed for coarse-grained visual tasks and may not transfer well to fine-grained domains such as plant species recognition. In this work, we investigate SSL for plant image representation learning. We show that commonly used augmentations in SSL pipelines - such as Gaussian blur, grayscale conversion, and solarization - are detrimental in the context of plant images, as they remove subtle discriminative cues essential for fine-grained recognition. We instead identify alternative transformations, including affine and posterization, that are better suited to this domain. We further demonstrate that training SimDINOv2 on the iNaturalist 2021 Plantae subset yields significantly stronger representations than training on ImageNet-1K, highlighting the importance of domain-specific data for SSL. Our findings are consistent across both ViT-Base and ViT-Large architectures. Moreover, our models achieve competitive performance and sometimes outperform strong supervised baselines Pl@ntCLEF and BioCLIP on downstream plant recognition tasks in few-shot settings. Overall, our results highlight the critical importance of domain-adapted augmentation strategies and dataset selection in self-supervised learning, and provide practical guidelines for building scalable models for biodiversity monitoring.

Problem

• self-supervised learning
• plant recognition
• fine-grained classification
• data augmentation
• domain adaptation

Innovation

• self-supervised learning
• domain-specific augmentation
• fine-grained plant recognition
• SimDINOv2
• biodiversity monitoring