Evaluasi Kompleksitas Algoritma Computer Vision Untuk Deteksi Objek Pada Aplikasi Informatika Terapan

Abstract

In recent years, object detection has become a crucial component in various computer vision applications, including autonomous driving, surveillance systems, and image recognition. The background of this research is driven by the growing need for efficient and accurate object detection methods capable of operating in real-time under hardware resource constraints. This study aims to provide a comprehensive comparative analysis of three prominent object detection algorithms: You Only Look Once (YOLO), Single Shot MultiBox Detector (SSD), and Faster Region-Based Convolutional Neural Networks (Faster R-CNN). The research methodology involves implementing these algorithms on standardized datasets, specifically the COCO dataset, and evaluating their performance based on various metrics including detection accuracy (mAP), processing speed (FPS/latency), and computational resource requirements. The main results indicate that YOLO excels in real-time applications due to its high speed, whereas Faster R-CNN demonstrates superior detection accuracy for precision-critical scenarios despite its slower processing performance. SSD offers a middle ground with respectable accuracy and speed, making it suitable for applications requiring a balance of both. In conclusion, this analysis highlights the unique strengths and weaknesses of each algorithm, providing valuable insights for researchers and practitioners in selecting the most appropriate object detection method based on specific application needs, whether prioritizing absolute precision or rapid real-time response.