Electronic Theses and Dissertation

Penelitian ini bertujuan mengimplementasikan sistem people line counting berbasis yolov8 yang terintegrasi dengan platform vms nx meta untuk mengatasi ketidakakuratan penghitungan pengunjung manual di perpustakaan universitas syiah kuala. metodologi penelitian mencakup akuisisi data video cctv, fine-tuning model yolov8, dan penerapan logika pelacakan objek menggunakan cross product analysis untuk mendeteksi arus masuk dan keluar pengunjung secara real-time. integrasi sistem dilakukan melalui metadata sdk guna mengirimkan data deteksi dalam format metadata terstruktur yang lebih efisien terhadap beban trafik jaringan dibandingkan transmisi video mentah. performa sistem diukur berdasarkan akurasi penghitungan arus, nilai mean average precision (map), efisiensi penggunaan bandwidth, serta stabilitas latensi respons sistem. luaran yang diharapkan adalah instrumen pemantauan otomatis pada antarmuka vms yang mampu menyajikan data statistik okupansi dengan performa tinggi dan presisi, sehingga memudahkan manajemen perpustakaan dalam menganalisis pola kepadatan pengunjung serta mendukung pengambilan keputusan berbasis bukti digital yang akurat. kata kunci: people line counting, yolov8, nx meta, metadata sdk, okupansi perpustakaan.

This study aims to implement a YOLOv8-based People Line Counting system integrated with the NX Meta VMS platform to address the inaccuracies of manual visitor counting at the Syiah Kuala University Library. The research methodology includes CCTV video data acquisition, fine-tuning the YOLOv8 model, and applying object tracking logic using Cross Product Analysis to detect visitor inflows and outflows in real-time. System integration is performed via the Metadata SDK to transmit detection data in a structured metadata format that is more efficient in terms of network traffic load compared to raw video transmission. System performance is measured based on flow counting accuracy, Mean Average Precision (mAP) values, bandwidth efficiency, and system response latency stability. The expected output is an automated monitoring tool on the VMS interface capable of presenting high-performance, precise occupancy statistics, thereby facilitating library management in analyzing visitor density patterns and supporting accurate, evidence-based decision-making. Keywords: People Line Counting, YOLOv8, NX Meta, Metadata SDK, Library Occupancy.