Robust Machine Learning Approach to Plant Species Classification for Sustainable Agriculture
Sara Mumtaz, Ahmad Jalal
Abstract
Sustaining resilient agricultural practices is essential for the overall yield because it guarantees both economic stability and food supplies. By incorporating Zero-Shot learning into a strong leaf classification architecture, our method meets this demand. We optimized an AI-powered four-step categorization process using the Bee Colony Algorithm. To pre-process, segment, extract features from, and classify leaf image data, zero-shot learning is employed. Pre-processing, which includes histogram equalization and contrast stretching, is followed by morphological segmentation. In feature extraction, two methods are used: Harris corner detection and SURF (Speeded Up Robust Features). Zero-Shot learning for classification is optimized using the Bee Colony Algorithm. At 92.73 % accuracy, the suggested method surpasses out previous methods like CNN and SVM, suggesting that it has the potential to enhance crop quality and disease control in agriculture.