Abstract:
The maturity level of oil palm Fresh Fruit Bunches (FFB) is a determining factor for the
quality of Crude Palm Oil (CPO). The sorting method after harvest or before entering
the boiling process is generally done manually relying on vision and experience. This
method is prone to subjective errors. Imaging methods are developing rapidly due to
advances in computers and image processing techniques, particularly for sorting and
grading systems. CenterNet is an architecture for object detection that is significantly
enhanced with the help of deep learning, CenterNet is a method for exploring the center
of a geometric grid. Plus a keypoint. This study used 18,000 FFB image data, the model
can detect the maturity level of oil palm from five maturity levels, namely ripe 1 in the
black-red color category, ripe 2 in the red color category, ripe 3 in the red-yellow color
category, in the yellow-orange category, and tanko with gray color category. The
performance of the results of the model has an accuracy rate of 88%.
