Sistem Deteksi dan Klasifikasi Jenis Kendaraan berbasis Citra dengan menggunakan Metode Faster-RCNN pada Raspberry Pi 4B

Mela Tri Audina (Fakultas Ilmu Komputer, Universitas Brawijaya)
Fitri Utaminingrum (Fakultas Ilmu Komputer, Universitas Brawijaya)
Dahnial Syauqy (Fakultas Ilmu Komputer, Universitas Brawijaya)

Jurnal Pengembangan Teknologi Informasi dan Ilmu Komputer
Vol 5 No 2 (2021): Februari 2021

Vehicles that exceed road capacity will have a negative impact on their surroundings, one of which causes accidents. Examples of cases of accidents that often occur are vehicles traveling in lanes that are not supposed to be, such as vehicles other than the busway crossing the busway lane and when driving on the Pantura highway which has more than 2 lanes, sometimes drivers find it very difficult to pay attention to the lane on the left, if you want. overtaking the vehicle in front of him. Therefore a system is needed to notify drivers to be more careful when driving. In this system there is a notification if there are numbers and types of vehicles in front. This system uses the Faster Regional Convolutional Neural Network modeling made on Tensorflow by processing it on a mini computer or Raspberry Pi 4B. The accuracy result in this system is 0.9025 or 90.25% with an average computation time in the Raspberry Pi 4B of 7,638 seconds per image.