Real-time object detection has become a fundamental component of modern computer vision, enabling intelligent systems to perceive and interpret dynamic environments across a wide range of applications. From autonomous transportation and video surveillance to medical imaging, robotics, and smart manufacturing, the demand for accurate and computationally efficient detection models continues to increase. Among deep learning-based approaches, the Region-Based Convolutional Neural Network (R-CNN) family and the You Only Look Once (YOLO) family represent two influential paradigms that have significantly shaped the evolution of object detection research. While R-CNN-based methods emphasize detection accuracy through region proposal mechanisms, YOLO architectures adopt a unified detection strategy that prioritizes real-time performance and deployment efficiency. This review presents a comprehensive comparative analysis of YOLO and R-CNN frameworks by examining their architectural developments, detection strategies, computational characteristics, and practical applicability. The study systematically reviews the progression from R-CNN to Fast R-CNN, Faster R-CNN, and Mask R-CNN, alongside the evolution of YOLO from its initial design to recent lightweight and high-performance variants. Key evaluation aspects, including detection accuracy, inference speed, computational complexity, memory requirements, robustness to challenging scenarios, and suitability for edge devices, are critically discussed. Furthermore, the paper highlights representative application domains, identifies existing research gaps, and examines emerging trends such as transformer-assisted detection, model compression, multimodal learning, and edge intelligence. The comparative findings indicate that the choice between YOLO and R-CNN architectures depends largely on application-specific requirements, balancing precision, latency, and resource constraints. This review provides researchers and practitioners with an organized perspective on current advancements and future directions for designing efficient and scalable real-time vision systems.

Real-time object detection, YOLO, R-CNN, Fast R-CNN, Faster R-CNN, Mask R-CNN, deep learning, computer vision, efficient vision systems, edge intelligence