An Interpretable Neural Network for Vegetation Phenotyping with Visualization of Trait-Based Spectral Features

TL;DR

This paper addresses vegetation phenotyping and species identification from hyperspectral VNIR-SWIR data using an interpretable neural network trained on the UPWINS spectral library. By comparing 26 ML methods and focusing on weight-visualization capable networks, it shows that learned spectral features correspond to chemical indicators (e.g., chlorophyll composition), enabling explanations beyond black-box predictions. The neural network achieves approximately 0.87 accuracy on a test set (with LDA around 0.91), complemented by Spectral Activation Plots and layer-1 weight visualizations that map spectral bands to class decisions, providing interpretable insight into vegetation traits. The work offers a framework for interpretable spectral-model design and open data sharing, with potential to extend to hyperspectral imagery and broader plant-phenotyping applications.

Abstract

Plant phenotyping is the assessment of a plant's traits and plant identification is the process of determining the category such as genus and species. In this paper we present an interpretable neural network trained on the UPWINS spectral library which contains spectra with rich metadata across variation in species, health, growth stage, annual variation, and environmental conditions for 13 selected indicator species and natural common background species. We show that the neurons in the network learn spectral indicators for chemical and physiological traits through visualization of the network weights, and we show how these traits are combined by the network for species identification with an accuracy around 90% on a test set. While neural networks are often perceived as `black box' classifiers, our work shows that they can be in fact more explainable and informative than other machine learning methods. We show that the neurons learn fundamental traits about the vegetation, for example the composition of different types of chlorophyll present which indicates species as well as response to illumination conditions. There is clear excess training capacity in our network, and we expect that as the UPWINS spectral library continues to grow the approach in this paper will provide further foundational insights in understanding plant traits. This provides a methodology for designing and interpreting neural networks on spectral data in general, and provides a framework for using neural networks with hyperspectral imagery for understanding vegetation that is extendable to other domains.

Figures (14)

• Figure 1: Counts of the classes in the current UPWINS spectral library.
• Figure 2: The mean spectrum for each class.
• Figure 3: The accuracy of the neural network on the training data during the training process.
• Figure 4: The confusion matrix for our neural network classification of the test subset.
• Figure 5: The matrix $W^1$ of the weights in layer 1 of the neural network.