<![CDATA[Mango (Mangifera indica L.) is one of the most important tropical fruits with high nutritional value and significant economic potential. However, manual identification of mango seedlings remains less accurate due to the similarities in leaf shape and size among different varieties, which often leads to misclassification. This study aims to develop an automated system to recognize five types of mango seedlings—Harum Manis, Indramayu, Golek, Madu, and Gedong Gincu by utilizing leaf vein textures as the main distinguishing features. The methodology employed the Local Binary Pattern (LBP) technique for feature extraction and a Backpropagation Neural Network (BPNN) as the classification model. The dataset consisted of 250 training images and 125 testing images with a resolution of 100×100 pixels, captured under varying lighting conditions ranging from one to five lamps. The experimental results indicate that lighting conditions significantly affect classification accuracy. The highest accuracy was achieved under four-lamp lighting conditions, reaching 91.20%, followed by two lamps (89.60%), three lamps (87.20%), five lamps (76.80%), and one lamp (67.20%). Furthermore, a BPNN configuration with 12 hidden neurons consistently demonstrated reliable recognition performance. These findings suggest that the combination of LBP and BPNN is effective for automatic classification of mango seedlings. The implementation of this system has the potential to assist farmers and seedling institutions by improving efficiency, accuracy, and reliability in seedling identification, thereby supporting the advancement of technology-based agriculture.]]>
Copyrights © 2025
