ANOTASI CITRA MULTISPEKTRAL UNTUK PEMBANGUNAN DATASET MODEL DETEKSI OBJEK PADA KEMATANGAN TANDAN BUAH SEGAR KELAPA SAWIT

Abstract

Computer vision as an object detection method can be used as a basis for automation process of sorting and grading oil palm FFB (Fresh Fruit Bunches). Images annotation are necessary to build datasets hence object detection model can identify object features in images. This study aims to annotate multispectral images of oil palm FFB with 2 categories, namely ripe and unripe FFB. Image acquisition is carried out with LED-based multispectral imaging system. Annotation is done using the python program application, called Labelimg. The number of annotated image data is 60 images consist of 30 images of ripe FFB and 30 images of unripe FFB. The results of annotation are .txt files contained information about bounding box coordinate, class object, and image dimension.