Deteksi dan Klasifikasi Kendaraan menggunakan Algoritma Backpropagation dan Sobel

Abstract

Abstract. Vehicle object recognition is a very interesting and everlasting research topic. The use of appropriate algorithms have a significant impact in the process of recognition and calculation. Proposed approach is designed and implemented using java. This system contain an application which can classify vehicles to recognize passing the toll road Surabaya-Malang km 34 using Backpropagation for learning and Sobel in Introduction to objects. In the process of image processing, the binary data of 10x10 pixels is used as input values for the neural network with 3 layers,and or leraning rate 0.3. Training process stops with a maximum value of 10,000 MSE (Mean Square Error) 0.0001. Success rate on the morning, noon, night test where 94.63%, 93.85% and 68.32%.