Browsing by Author "Wardana, Fiqra"

Now showing 1 - 2 of 2
  • No Thumbnail Available
    Item
    SISTEM DETEKSI TINGKAT KEMATANGAN TANDAN BUAH SEGAR KELAPA SAWIT MENGGUNAKAN METODE CONVNET BERBASIS ANDROID
    (Elfitra, 2022-06) Wardana, Fiqra; Bahri, Zaiful
    One of the oil palms harvesting processes is to determine the maturity level of Fresh Fruit Bunches (FFB). FFB maturity is one of the determinants quality productions of palm oil processing materials. In general, FFB maturity can be checked manually by farmers by direct observation. Manual selection of FFB, of course, requires time and experienced farmers to be able to determine maturity correctly. ConvNet is a machine learning method that can be used to quickly determine the maturity level of oil palm FFB. ConvNet allows the model to recognize the shape, color, and edge of each TBS training data. Using 900 FFB image data, the model can detect the maturity level of oil palm from three classes, namely raw, ripe, and empty bunches. The results of the training model have an accuracy of 86% with a precision and recall of more than 90%.
  • No Thumbnail Available
    Item
    SISTEM DETEKSI TINGKAT KEMATANGAN TANDAN BUAH SEGAR KELAPA SAWIT MENGGUNAKAN METODE CONVNET BERBASIS ANDROID
    (Elfitra, 2022-06) Wardana, Fiqra; Bahri, Zaiful
    One of the oil palms harvesting processes is to determine the maturity level of Fresh Fruit Bunches (FFB). FFB maturity is one of the determinants quality productions of palm oil processing materials. In general, FFB maturity can be checked manually by farmers by direct observation. Manual selection of FFB, of course, requires time and experienced farmers to be able to determine maturity correctly. ConvNet is a machine learning method that can be used to quickly determine the maturity level of oil palm FFB. ConvNet allows the model to recognize the shape, color, and edge of each TBS training data. Using 900 FFB image data, the model can detect the maturity level of oil palm from three classes, namely raw, ripe, and empty bunches. The results of the training model have an accuracy of 86% with a precision and recall of more than 90%.