IMPLEMENTASI METODE CNN UNTUK KLASIFIKASI OBJEK

Abstract

Objects can be interpreted as all inanimate and living things that have various shapes and sizes. For humans to determine the presence of objects, to classify and estimate the distance of objects around them is not difficult. But for a computer to do the work mentioned above with an accuracy level that reaches up to greater than 90% is not easy. Object detection is important in the field of computer vision because it is used to monitor and track objects, while robots that use cameras as sensors are used to avoid obstacles, follow objects, classify and so on. Therefore, the purpose of this study was to determine the level of accuracy of the CNN method in classifying objects. The steps used to complete this research were literature study, collecting digital image data, determining training, and testing data, designing the CNN program, conducting training and testing. From the results of testing the CNN method that has been carried out, it is known that the level of accuracy in classifying objects reaches 98%.