Energy-Efficient Plant Monitoring via Knowledge Distillation

arXiv cs.CV / 5/1/2026

📰 NewsIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper addresses the high compute cost of modern, large visual models for plant species and plant disease recognition in resource-constrained settings like mobile and edge devices.
It studies knowledge distillation to transfer representational power from large pretrained models into smaller, more efficient architectures.
Experiments cover four representative model families (two ConvNeXt and two vision transformers) trained under multiple regimes, including from-scratch vs pretrained initialization, with and without distillation.
Across two challenging benchmarks (Pl@ntNet300K-v2 and Deep-Plant-Disease), the authors find that knowledge distillation consistently improves performance and helps compact “student” models match much larger models while using far less compute.
The results suggest knowledge distillation can make automated biodiversity monitoring and precision agriculture more scalable and practical in real-world environments.

Abstract

Recent advances in large-scale visual representation learning have significantly improved performance in plant species and plant disease recognition tasks. However, state-of-the-art models, often based on high-capacity vision transformers or multimodal foundation models, remain computationally expensive and difficult to deploy in resource-constrained environments such as mobile or edge devices. This limitation hinders the scalability of automated biodiversity monitoring and precision agriculture systems, where efficiency is as critical as accuracy. In this work, we investigate knowledge distillation as an effective approach to transfer the representational capacity of large pretrained models into smaller, more efficient architectures. We focus on plant species and disease recognition, and conduct an extensive empirical study on two challenging benchmarks: Pl@ntNet300K-v2 and Deep-Plant-Disease. We evaluate four representative architectures, including two ConvNeXt models and two vision transformers, under multiple training regimes: from-scratch training and pretrained initialization, each with and without distillation. In total, we train and evaluate 70 models. Our results show that knowledge distillation consistently improves performance across tasks and architectures. Distilled models are able to match the performance of significantly larger models while maintaining substantially lower computational cost. These findings demonstrate the potential of knowledge distillation techniques to enable efficient and scalable deployment of plant recognition systems in real-world environmental applications.