Utilization of Computer vision for onion (Allium cepa) disease identification
No Thumbnail Available
Views
...
Downloads
...
Date
2024
Authors
Leo P. Sarmiento and Joyce Ann S. Solomon
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Onion (Allium cepa L.) locally referred to as “sibuyas”, stands as pivotal culinary ingredient, playing a fundamental role in the quintessential seasoning of dishes. Within the Philippines, the cultivation of onions has thrived, positioning it as one of the top-performing crops and a prized agricultural commodity. Despite its paramount importance, the onion crop grapples with vulnerability to an array of plant diseases, posing a considerable hindrance to the optimal production of onions in the region. This susceptibility underscores the need for strategic interventions and research to safeguard and enhance onion yields in the face of these agricultural challenges.
The study considers the use of the camera as sensor, incorporating elements of computer vision and machine learning. The primary focus is on leveraging these technologies to identify a spectrum of diseases commonly affecting onions. These diseases encompass Botrytis Leaf Blight, Downy Mildew, Purple Blotch, Onion Rust, Stemphylium Leaf Blight and Xanthomonas Leaf Blight.
The result shows the overall accuracy of the developed system is 94% indicating that the model performs well in general. The Precision, Recall and F-1 Score for each class are in the range of 0.89 to 0.96. In terms of model training and validation, the training loss decreases from 0.8 to 0.1, and the validation loss decreases from 0.9 to 0.2 over the epoch.
This indicates an improvement in the model’s performance on the training data ang suggesting generalization to unseen data, In terms of calibration, the model prediction probability is less than the true outcome indicating good calibration.