Claim Missing Document
Check
Articles

Found 2 Documents
Search

Object Distance Measurement System Using Monocular Camera on Vehicle Fussy Mentari Dirgantara; Arief Syaichu-Rohman; Lenni Yulianti
Proceeding of the Electrical Engineering Computer Science and Informatics Vol 6: EECSI 2019
Publisher : IAES Indonesia Section

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eecsi.v6.1953

Abstract

To support autonomous vehicles that are currently often studied by various parties, the authors propose to make a system of predicting the distance of objects using monocular cameras on vehicles. Distance prediction uses four methods and the input parameter was obtained from images processed with MobileNets SSD. Calculations using linear regression are the simplest calculations among the four methods but have an error of 1% with a standard deviation of 1.65 meters. While using the first method, the average error value is 9% with a standard deviation of 0.43 meters. By using the second calculation, the average error resulted in 6% with a standard deviation of 0.35 meters. The experimental method had an average error of 1% with a standard deviation of 0.26 meters, so the experimental method was used.
Kombinasi Deteksi Objek, Pengenalan Wajah dan Perilaku Anomali menggunakan State Machine untuk Kamera Pengawas MUHAMMAD FAUZI NURYASIN; CARMADI MACHBUB; LENNI YULIANTI
Jurnal Elkomika Vol 11, No 1 (2023): ELKOMIKA: Jurnal Teknik Energi Elektrik, Teknik Telekomunikasi, & Teknik Elektr
Publisher : Institut Teknologi Nasional, Bandung

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.26760/elkomika.v11i1.86

Abstract

ABSTRAKSaat ini sistem kamera pengawas mengandalkan manusia dalam melakukan penerjemahan pada rekaman gambar yang terjadi. Perkembangan computer vision, machine learning, dan pengolahan citra dapat dimanfaatkan untuk membantu peran manusia dalam melakukan pengawasan. Penelitian ini merancang sistem kerja kamera yang terdiri dari tiga modul yaitu deteksi objek, pengenalan wajah, dan perilaku anomali. Deteksi objek memakai HOG-SVM, pengenalan wajah menggunakan CNN dengan arsitektur VGG-16 memanfaatkan transfer learning, dan perilaku anomali memakai spatiotemporal autoencoder berdasarkan threshold. Ketiga modul tersebut diuji menggunakan metrik akurasi, presisi, recall, dan f1-score. Ketiga modul diintegrasikan dengan state machine menjadi satu kesatuan sistem. Kinerja modul memiliki akurasi 88% untuk deteksi objek, 98% untuk pengenalan wajah, dan 78% untuk perilaku anomali. Hasil tampilan riil dapat diakses secara sederhana dan nirkabel melalui web.Kata kunci: HOG-SVM, CNN, VGG-16, spatiotemporal autoencoder, state machineABSTRACTNowadays, the surveillance camera system relies on human to interpret the recorded images. Computer vision, machine learning, and image processing can be utilized to assist the human role in supervising. This study designed a camera work system consisting of three main modules, namely object detection, face recognition, and anomaly behavior. Object detection used the HOG-SVM combination. Facial recognition used CNN with the VGG-16 architecture that utilized transfer learning. Anomalous behavior used spatiotemporal autoencoder based on threshold. Modules are tested using the metrics of accuracy, precision, recall, and f1-score. The three modules are integrated using a state machine into one system. The performance of the module had 88% accuracy for object detection, 98% for facial recognition, and 78% for anomalous behavior. Real time video recording can be accessed wireless via web-based.Keywords: HOG-SVM, CNN, VGG-16, spatiotemporal autoencoder, state machine