<![CDATA[Detecting the ripeness of oil palm fruit bunches is a crucial process in the palm oil industry to ensure the quality and quantity of oil extracted. Conventional methods still rely on subjective and inefficient manual observation. This study proposes a visual detection system using the Simple Evolving Connectionist System (SECoS) algorithm to identify the ripeness of oil palm bunches based on visual images. This model utilizes color, texture, and shape characteristics extracted from images and processed through an adaptive and evolving neural network structure. The results demonstrate that SECoS is capable of high detection accuracy and adapts to new data patterns. This system has the potential to be applied in precision agriculture practices. The model achieved an average accuracy of 91.3%, with the highest accuracy of 94% in the "Ripe" category in the final test based on 300 dataset. This demonstrates that parameter optimization is crucial in improving the model's ability to adapt to variations in oil palm bunch image data. Accuracy improvements were evident in both training and validation data. However, not all categories achieved optimal results, with accuracy for the "empty bunch" labels (89%) and "unripe" labels (88%) being relatively lower than for the other categories.]]>
Copyrights © 2026
