[Paper Review] A comprehensive review on Plant Leaf Disease detection using Deep learning

Sumaya Mustofa, Md Mehedi Hasan Munna+3 more|arXiv (Cornell University)|Aug 27, 2023

Smart Agriculture and AI15 citations

TL;DR

This paper delivers a systematic review of deep learning models for plant leaf disease detection, comparing architectures like ViT, DCNN/CNN, RSNSR-LDD, DDN, and YOLO across public datasets using common metrics.

ABSTRACT

Leaf disease is a common fatal disease for plants. Early diagnosis and detection is necessary in order to improve the prognosis of leaf diseases affecting plant. For predicting leaf disease, several automated systems have already been developed using different plant pathology imaging modalities. This paper provides a systematic review of the literature on leaf disease-based models for the diagnosis of various plant leaf diseases via deep learning. The advantages and limitations of different deep learning models including Vision Transformer (ViT), Deep convolutional neural network (DCNN), Convolutional neural network (CNN), Residual Skip Network-based Super-Resolution for Leaf Disease Detection (RSNSR-LDD), Disease Detection Network (DDN), and YOLO (You only look once) are described in this review. The review also shows that the studies related to leaf disease detection applied different deep learning models to a number of publicly available datasets. For comparing the performance of the models, different metrics such as accuracy, precision, recall, etc. were used in the existing studies.

Motivation & Objective

• Survey the landscape of deep learning models used for plant leaf disease detection.
• Summarize advantages and limitations of different architectures for leaf disease tasks.
• Discuss datasets and evaluation metrics employed in leaf disease studies.
• Highlight gaps and future directions in the field.

Proposed method

• Systematic literature review of leaf disease detection studies using deep learning.
• Classification and object-detection architectures analyzed (ViT, CNN, RSNSR-LDD, DDN, YOLO).
• Comparison based on reported metrics such as accuracy, precision, and recall across public datasets.

Experimental results

Research questions

• RQ1 What deep learning architectures have been applied to plant leaf disease detection?
• RQ2 What are the reported performance metrics and datasets used in these studies?
• RQ3 What are the main advantages and limitations of each model type for leaf disease tasks?
• RQ4 What gaps exist in current leaf disease detection research and how can they be addressed?

Key findings

• Multiple deep learning models have been applied to leaf disease detection, including ViT, CNN-based networks, RSNSR-LDD, DDN, and YOLO.
• Studies report using various public datasets and common metrics such as accuracy, precision, and recall to evaluate performance.
• The review discusses advantages and limitations of each model category in the context of leaf disease detection.
• There is variability in datasets and evaluation protocols, indicating a need for standardized benchmarks.