Evaluating Histogram Matching for Robust Deep learning-Based Grapevine Disease Detection

Ruben Pascual; Inés Hernández; Salvador Gutiérrez; Javier Tardaguila; Pedro Melo-Pinto; Daniel Paternain; Mikel Galar

cs.CVApr 21, 2026

Evaluating Histogram Matching for Robust Deep learning-Based Grapevine Disease Detection

Authors: Ruben Pascual, Inés Hernández, Salvador Gutiérrez, Javier Tardaguila, Pedro Melo-Pinto, Daniel Paternain, Mikel Galar

Organizations: Institute of Smart Cities (ISC), Dept. of Statistics, Computer Science and Mathematics, Public Univ. of Navarre (UPNA), 31006 Pamplona, Spain · Institute of Grapevine and Wine Sciences (Univ. of La Rioja, CSIC, Govt. of La Rioja), 26007, Logroño, Spain and Televitis Research Group, Univ. of La Rioja, 26006, Logroño, Spain · Dept. of Computer Science and AI, Univ. of Granada, 18071, Granada, Spain · Centre for the Research and Technology of Agroenvironmental and Biological Sciences (CITAB), Inov4Agro, and Departamento de Engenharias, UTAD, Quinta Dos Prados, 5000-801, Vila Real, Portugal

Abstract

Variability in illumination is a primary factor limiting deep learning robustness for field-based plant disease detection. This study evaluates Histogram Matching (HM), a technique that transforms the pixel intensity distribution of an image to match a reference profile, to mitigate this in grapevine classification, distinguishing among healthy leaves, downy mildew, and spider mite damage. We propose a dual-stage integration of HM: (i) as a preprocessing step for normalization, and (ii) as a data augmentation technique to introduce controlled training variability. Experiments using 1,469 RGB images (comprising homogeneous leaf-focused and heterogeneous canopy samples) to train ResNet-18 models demonstrate that this combination significantly enhances robustness on real-world canopy images. While leaf-focused samples showed marginal gains, the canopy subset improved markedly, indicating that balancing normalization with histogram-based diversification effectively bridges the domain gap caused by uncontrolled lighting.

Explore similar work

Towards Robust Deep Learning-based Rumex Obtusifolius Detection from Drone Images
Apr 28, 2026Fabian Dionys Schrag, Mehmet Ozgur Turkoglu, Konrad Schindler +1Domain Adaptation Aerial Dataset
Few-Shot Hyperspectral Aphid Detection via FastGAN Synthetic Data Generation, Transformer-Based Classification and Explainable AI
Jun 19, 2026Ali SaeidanHyperspectral Image Classification Tomato Diseases
Learning Adversarial Augmentation Policies for Robust Garlic Seedling Detection
Jun 25, 2026Soeun Lee, Chanho Kim, Yeji Kang +2Tomato Diseases Aerial Imagery

Apr 28, 2026 · cs.CVJ/K move · Enter open · S save

Towards Robust Deep Learning-based Rumex Obtusifolius Detection from Drone Images

Fabian Dionys Schrag, Mehmet Ozgur Turkoglu, Konrad Schindler, Ralph Lukas Stoop

Agroscope NBA, Tänikon 1, Ettenhausen, 8356, TG, Switzerland · Agroscope Earth Observation of Agroecosystems Team, Reckenholz 191, Zurich, 8046, ZH, Switzerland · ETH Zurich, Stefano-Franscini-Platz 5, Zurich, 8093, ZH, Switzerland

Domain adaptation (DA) addresses the challenge of transferring a machine learning model trained on a source domain to a target domain with a different data distribution. In this work, we study DA for the task of Rumex obtusifolius (Rumex) image classification. We train models on a published, ground vehicle-based dataset (source) and evaluate their performance on a custom target dataset acquired by unmanned aerial vehicles (UAVs). We find that Convolutional Neural Network (CNN) models, specifically ResNets, generalize poorly to the target domain, even after fine-tuning on the source data. Applying moment-matching and maximum classifier discrepancy, two established DA techniques, substantially improves target-domain performance. However, Vision Transformer (ViT) models pretrained with self-supervised objectives (DINOv2, DINOv3) handle domain shifts intrinsically well, surpassing even moment-matching-trained ResNets, likely due to the rich, general-purpose representations acquired during large-scale pretraining. Using ViTs fine-tuned on the source dataset, we demonstrate high classification performances in the range of F1=0.8 on our target dataset. To support further research on DA for weed detection in grassland systems, we publicly release our UAV-based target dataset AGSMultiRumex, comprising data from 15 flights over Swiss meadows.

Domain Adaptation Aerial Dataset Image Classification Target Domain Residual Connections

Open paper Ask about it PDF

Abstract

Explore similar work

Towards Robust Deep Learning-based Rumex Obtusifolius Detection from Drone Images

Few-Shot Hyperspectral Aphid Detection via FastGAN Synthetic Data Generation, Transformer-Based Classification and Explainable AI

Learning Adversarial Augmentation Policies for Robust Garlic Seedling Detection