CD Skripsi
Rancang Bangun Sistem Penghitung Jumlah Tandan Buah Segar (Tbs) Kelapa Sawit Pada Sistem Konveyor Berjalan Berbasis Citra
FFB (Fresh Fruit Bunches) of oil palm is one of the indonesian commodities that
make it possible for Indonesia to be country of the largest CPO (Crude Palm Oil)
export in the world. Automatic counting of oil palm FFB is very important in the
sorting and grading process to calculate yields, estimate the amount of fruit storage
material, and estimate the storage facility. That is very crucial for efficiency in time,
cost, and labour. This study aimed to design a program due to automatically count
moving oil palm FFB using computer vision in video frames. The system is equipped
with a webcam, light source, conveyor and software. The counting program consists
of two main components, a detector and a tracker combined with a counter. The
detection of oil palm FFB based on saturation of color parameters. Output this
process is location of an oil palm FFB in a bounding box. The tracker will track the
bounding box of object in each video frame and the counter will count the bounding
boxes passes the camera. System will be given a specific ID with iteration system.
Video of oil palm FFB on a moving conveyor was recorded using webcam with
frame rate of 20. Samples were 117 oil palm FFBs of Tenera varieties which recorded
using 10 videos. The counting results were validated using method of confusion
matrix. The results show that the counting accuracy depend on room light intensity
due to colour segmentations using Hue, Saturation, Value (HSV) method. The
confusion matrix validation resulted in average counting accuracy of 79.25 %.
Keywords: Confusion matrix, computer vision, oil palm fresh fruit bunch, video
frame, counting fruit.
Tidak tersedia versi lain