A comparative study of machine learning models for plant disease identification
Măcelaru Mara, Petrică Pop, Jose Barata

Abstract. Plant disease identification from images is a central task in precision agriculture, but the rapid evolution of machine learning makes it difficult to determine which model class is most appropriate for a given setting. This paper presents a comparative study of several machine learning and deep learning models for plant disease identification, evaluated on common benchmark datasets. The compared models include classical machine learning approaches based on engineered features, convolutional neural networks of different depths and pre-trained transfer-learning architectures. Accuracy, robustness, computational cost and suitability for deployment on edge devices are discussed, providing practical guidance for designing plant disease detection pipelines in smart agriculture.

Keywords: plant disease identification; machine learning; deep learning; image classification; precision agriculture

Măcelaru Mara, Petrică Pop, Jose Barata, "A comparative study of machine learning models for plant disease identification", Proc. 19th SOCO Int. Conf. on Soft Computing Models in Industrial and Environmental Applications, Springer, 2024, 2023.

See also: An Enhanced Hybrid Machine Learning Model for Plant Disease Detection and Classification (HAIS 2025) — a follow-up work by the same lead author with improved hybrid approach.

Reference: Proc. 19th SOCO Int. Conf. on Soft Computing Models in Industrial and Environmental Applications, Springer, 2024.

Other publications