EVALUATION OF HISTOGRAM OF ORIENTED GRADIENT (HOG) AND LEARNING VECTOR ALGORITHM QUANTIZATION (LVQ) IN CLASSIFICATION CARICA VASCONCELLEA CUNDINAMARCENCIS

Abstract

Papaya fruit has various types, one of which is carica. Carica is another name for a papaya which grows in mountain areas, such as Dieng and Wonosobo. The fruit, which when ripe, will be a breech ovoid and has a size of about 6–15 cm x 3–8 cm. So, classification is needed to differentiate carica maturity levels so that in the processing process producers can easily get good quality Carica fruit. The method used as research is the LVQ method to classify carica fruits. And HOG (Histogram of Oriented Gradients) algorithm is one method for extracting features from image objects. The process of HOG is to convert an RGB (Red, Green Blue) image to grayscale. Following are the steps of the HOG algorithm process: Image Conversion, Gradient Computer, Spatial Orientation Binning, Block, and Windows Detector. From the results of extracting HOG on training and testing data, there are some conclusions including; 104 datasets consisting of 80 training data and 24 testing data that were applied to classify the maturity level of carica fruit using the HOG and LVQ methods can produce the highest accuracy rate of 91,67% with HOG size values are 256, 128, and 64. And the size of HOG processes is 256 with the fastest time which is equal to 38.52 seconds. According to research conducted by the HOG and LVQ methods, it can and successfully be applied to the image of carica fruit to classify the maturity of the fruit.