MapReduce-based deep learning framework for potato leaf disease detection in sustainable precision agriculture.

Area of Science:

• Agricultural Science
• Computer Science
• Data Science

Background:

• Potato leaf diseases significantly threaten crop yield and global food security.
• Accurate and timely disease detection is crucial for minimizing yield losses in precision agriculture.

Purpose of the Study:

• To develop and evaluate a lightweight MobileNetV3 classifier integrated with a MapReduce pipeline for efficient potato leaf disease detection.
• To enhance crop productivity and sustainability in precision farming through automated disease monitoring.

Main Methods:

• A dataset of 2152 potato leaf images was used, categorized into three classes.
• A preprocessing pipeline involved image resizing, normalization, and data augmentation.
• A lightweight MobileNetV3 classifier was paired with a MapReduce-style data pipeline for parallel preprocessing and batch inference.

Main Results:

• The model achieved high accuracy: 98.6% (training), 96.9% (validation), and 96.8% (testing).
• Key performance metrics included testing sensitivity (95.3%), specificity (97.7%), and F1-Score (96.4%).
• The MapReduce pipeline demonstrated scalability for large datasets and continuous image ingest, outperforming single-node baselines in throughput.

Conclusions:

• The proposed MobileNetV3 and MapReduce system offers a scalable and robust solution for large-scale agricultural disease monitoring.
• This approach significantly enhances precision farming by enabling accurate and efficient potato disease detection.
• The model's high performance and extensibility support sustainable agriculture and global food security efforts.