Revolutionizing crop disease detection with computational deep learning: a comprehensive review.

Area of Science:

• Agricultural Science
• Computer Science
• Data Science

Background:

• Deep learning (DL) algorithms outperform conventional methods in crop detection and disease identification.
• DL applications translate plant images into actionable insights for early disease diagnosis.

Purpose of the Study:

• To provide a comprehensive review of contemporary literature on crop disease diagnosis, categorization, and severity assessment.
• To analyze the performance of machine learning (ML) and DL techniques in recent studies.
• To identify research gaps and provide recommendations for future investigations.

Main Methods:

• Review of contemporary literature on ML and DL for crop disease diagnosis.
• Performance analysis of various ML and DL techniques including CNN, KNN, SVM, and ANN.
• Scrutiny of methodologies, datasets, and research gaps.

Main Results:

• Most studies focus on traditional ML algorithms and CNN, with limited exploration of emerging DL algorithms like capsule networks and vision transformers.
• Existing datasets are often crop-specific, highlighting the need for diverse, comprehensive image datasets.
• Research predominantly addresses individual diseases or algorithms, rather than integrated approaches.

Conclusions:

• There is a need to explore emerging DL algorithms and develop comprehensive datasets for broader applicability.
• A unified framework combining ML and DL is advocated for effectively addressing multiple plant diseases.