Abstract:
Identifying the maturity level of Fresh Fruit Bunches is one of the steps in the oil palm
harvesting process. The maturity level of Fresh Fruit Bunches is an important factor that
determines the quality of palm oil production. In general, farmers can check the maturity
of FFB manually by direct observation. Determining the maturity of FFB manually
requires time and expertise of skilled farmers to ensure proper ripeness. YOLO v5 is a
machine learning method that can be used to quickly and accurately detect oil palm FFB
objects. With YOLO v5, the model can recognize the shape, color and edge of each FFB
training data. The software used is google collaboratory, android studio, roboflow,
tensorfow lite, and drawio. For UML system design using Drawio software using 8002
FFB image dataset consisting of 5600 training data, 1600 validation data and 799 test
data. The model can identify six levels of oil palm maturity consisting of raw, ripe 1, ripe
2, ripe 3, over ripe, and empty bunches. The results of the training model have an
accuracy of 82% with a precision, recall and an F1 score of more 94%.
