Automated Okra Maturity Detection Using Thermal Imaging and Deep Learning

Document Type

Book

Publication Date

2026

Abstract

The world faces critical challenges like food insecurity, malnutrition, and climate change, emphasizing the need for sustainable agricultural solutions. Okra, a nutrient-dense and resilient crop, holds promise by supporting Goal 2: Zero Hunger and Goal 12: Responsible Consumption and Production. With its adaptability to diverse climates and minimal environmental impact, okra enhances food security in resource-limited regions and promotes sustainable farming practices. Leveraging the benefits of okra could strengthen resilience and sustainability in global food systems, presenting an effective solution to pressing food and environmental issues. This chapter introduces a model for distinguishing between over-matured and adequately matured okra using an automated okra maturity detection model designed to provide high accuracy and reliability. The proposed automated okra maturity detection model consists of three main phases: data preprocessing, data classification, and model evaluation. In the preprocessing phase, challenges such as unwanted noise and class imbalance are addressed. Next, the custom Convolutional Neural Network (CNN) model is trained using the refined dataset from the first phase. Model performance is then assessed based on four metrics: accuracy, precision, recall, and F1-score. The proposed model achieved 95% accuracy, 95% precision, 94.5% recall, and 94.5% F1 score on the test set. When compared to existing models, the proposed model demonstrated superior performance, particularly in terms of average recall and average F1 score.

Link to Full Text

Share

COinS