| dc.contributor.advisor | Indriasari, Sofiyanti | |
| dc.contributor.author | Salsabila, Alya Putri | |
| dc.date.accessioned | 2026-07-03T03:35:41Z | |
| dc.date.available | 2026-07-03T03:35:41Z | |
| dc.date.issued | 2026 | |
| dc.identifier.uri | http://repository.ipb.ac.id/handle/123456789/173989 | |
| dc.description.abstract | Penelitian ini dilatarbelakangi oleh proses verifikasi presensi teknisi di PT Eka Mas Republik yang proses pelaksanaannya mengandalkan pemeriksaan manual dan dapat menyebabkan terjadinya human error serta ketidakefisienan waktu. Penelitian ini memiliki tujuan untuk melakukan pengembangan dan evaluasi terhadap sistem presensi berbasis deteksi atribut teknisi menggunakan model YOLOv8 dengan pendekatan CRISP-DM. Dataset yang digunakan sebanyak 731 citra yang mencakup 391 citra teknisi serta 340 citra non-teknisi, dengan proporsi 70% untuk data pelatihan sebanyak 511 citra, 20% untuk data validasi sebanyak 146 citra, dan 10% data pengujian sebanyak 74 citra. Performa model dianalisis berdasarkan nilai precision, recall, dan mean Average Precision (mAP@0.5), serta waktu inferensi untuk mengukur efisiensi sistem. Berdasarkan hasil penelitian menggunakan data test, model YOLOv8 menghasilkan nilai precision sebesar 0,974, recall sebesar 0,932, serta mAP@0.5 sebesar 0,959, dengan waktu inferensi rata-rata sebesar 11,0 ms per citra. Selain itu, YOLOv8 menunjukkan performa yang lebih optimal dibandingkan YOLOv5 dan YOLOv11. Sistem yang dikembangkan mampu mengotomatisasi proses verifikasi atribut teknisi dan meningkatkan efisiensi operasional dalam proses presensi. | |
| dc.description.abstract | Research was motivated by the technician attendance verification process at PT Eka Mas Republik, which depends on manual checking procedures that can result in human error and reduced time efficiency. This research aims to develop and evaluate an attendance system based on technician attribute detection using the YOLOv8 model with the CRISP-DM approach. The dataset consists of 731 images, including 391 images of technicians and 340 images of non-technicians, with a distribution of 70% for training data (511 images), 20% for validation data (146 images), and 10% for testing data (74 images). Model performance was analyzed based on precision, recall, and mean Average Precision (mAP@0.5), as well as inference time to measure system efficiency. Based on the research results, the YOLOv8 model achieved a precision of 0.974, a recall of 0.932, and a mAP@0.5 of 0.959, with an average inference time of 11.0 ms per image. Additionally, YOLOv8 demonstrated greater performance than YOLOv5 and YOLOv11. The developed system is capable of automating the technician attribute verification process and improving operational efficiency in the attendance process. | |
| dc.description.sponsorship | | |
| dc.language.iso | id | |
| dc.publisher | IPB University | id |
| dc.title | Pengembangan dan Evaluasi Sistem Presensi Berbasis YOLOv8 melalui Pendekatan CRISP-DM di PT Eka Mas Republik | id |
| dc.title.alternative | Development and Evaluation of a YOLOv8-Based Attendance System Using the CRISP-DM Approach at PT Eka Mas Republik | |
| dc.type | Tugas Akhir | |
| dc.subject.keyword | Attendance system | id |
| dc.subject.keyword | computer vision | id |
| dc.subject.keyword | CRISP-DM | id |
| dc.subject.keyword | technician attribute detection | id |
| dc.subject.keyword | YOLOv8 | id |
| dc.subtype | Undergraduate Theses | |