Implementasi You Only Look Once (YOLO) V3 untuk deteksi objek pada aktivitas manusia di ruangan kelas

Abdillah, Fajrian Fatan (2023) Implementasi You Only Look Once (YOLO) V3 untuk deteksi objek pada aktivitas manusia di ruangan kelas. Sarjana thesis, UIN Sunan Gunung Djati Bandung.

Preview	Text (COVER) 1_cover.pdf Download (97kB) | Preview
Preview	Text (ABSTRAK) 2_abstrak.pdf Download (142kB) | Preview
Preview	Text (DAFTA ISI) 3_daftarisi.pdf Download (136kB) | Preview
Preview	Text (BAB I) 4_bab1.pdf Download (214kB) | Preview
	Text (BAB II) 5_bab2.pdf Restricted to Registered users only Download (541kB) | Request a copy
	Text (BAB III) 6_bab3.pdf Restricted to Registered users only Download (387kB) | Request a copy
	Text (BAB IV) 7_bab4.pdf Restricted to Registered users only Download (486kB) | Request a copy
	Text (BAB V) 8_bab5.pdf Restricted to Registered users only Download (79kB) | Request a copy
	Text (DAFTAR PUSTAKA) 9_daftarpustaka.pdf Restricted to Registered users only Download (226kB) | Request a copy

Abstract

INDONESIA : Pada tahun 2021, Badan Pusat Statistik (BPS) menerbitkan buku berjudul “Statistik Kriminal 2021” yang berisi laporan tahunan jumlah kasus kejahatan yang terjadi di Indonesia. Dalam buku tersebut dinyatakan pada tahun 2020 pencurian tanpa kekerasan menduduki peringkat tertinggi dengan total 73.264 kasus, lalu berikutnya pencurian dengan kekerasan terjadi sebanyak 6.358 kasus. Hal ini menunjukkan bahwa kasus pencurian tanpa kekerasan masih sering terjadi, sehingga diperlukan sistem deteksi yang dapat mendukung keamanan lingkungan, termasuk ruangan kelas. Perangkat kamera Closed Circuit Television (CCTV) secara umum dapat mencegah pencurian jika sedang dipantau oleh penjaga keamanan, namun perangkat ini memiliki kelemahan yaitu sifatnya yang pasif. Tujuan penelitian ini adalah mengimplementasikan dan mengetahui tingkat akurasi You Only Look Once (YOLO) v3 untuk mendeteksi objek pada aktivitas manusia di ruangan kelas. Dataset penelitian adalah video rekaman aktivitas manusia di ruangan kelas di gedung Pusat Teknologi Informasi dan Pangkalan Data (PTIPD) Universitas Islam Negeri Sunan Gunung Djati Bandung. Penelitian dilakukan dengan metode Cross-Industry Standard Process for Data Mining (CRISP-DM) yang terdiri dari 5 skenario pengujian model YOLO v3. Hasil pengujian menunjukkan bahwa model YOLO v3 menghasilkan nilai akurasi sebesar 99.1% dari pengujian ke-5. ENGLISH : In 2021, the Central Statistics Agency (BPS) published a book entitled "Criminal Statistics 2021" which contains an annual report on the number of crime cases that have occurred in Indonesia. The book states that in 2020 non-violent theft has the highest ranking with a total of 73,264 cases, followed by theft with violence with 6,358 cases. This shows that cases of theft without violence are still common, so a detection system is needed that can support environmental security, including classrooms. Closed Circuit Television (CCTV) camera devices in general can prevent theft if they are being monitored by security guards, but this device has a weakness which is it have passive behavior. This study aimed to implement and determine the accuracy of You Only Look Once (YOLO) v3 for detect objects in human activities in the classroom. The research dataset is a video recording of human activity in a classroom at the Center for Information Technology and Data Base building, State Islamic University Sunan Gunung Djati, Bandung. The research was conducted using the Cross-Industry Standard Process for Data Mining (CRISP-DM) method which consisted of 5 test scenarios for the YOLO v3 model. The test results show that the YOLO v3 model produces an accuracy value of 99.1% from the 5th test.

Actions (login required)

View Item