Object Detection is a fundamental, but challenging problem in computer vision. However, the rise of deep learning techniques has led to creation of numerous detection frameworks that strike an optimal trade-off between speed and accuracy. These models are split into two categories: single stage and two-stage detectors. This study evaluates six prominent object detection models: four single-stage detectors (YOLOv3, YOLOv8n, YOLOv8x, and YOLOv11), and two two-stage detectors (Fast R-CNN and Faster R-CNN). Performance of these models is evaluated using mean Average Precision (mAP) and inference time. Two-stage methods, such as Fast R-CNN and Faster R-CNN, use region proposals to enhance detection accuracy, especially in visually complex scenes. Faster R-CNN was shown to achieve the highest detection accuracy with an mAP score of 85.72%, albeit with slower processing time. In comparison, Single-stage methods, or the variants of “YOLO” models are more focused on providing faster detection of more objects through less complex designs. Among them YOLOv8n was fastest and achieved highest mAP of 70.91%, followed by YOLOv3 (66.53%), YOLOv11(61.63%), and YOLOv8x(55.39%). These findings highlight that two-stage detectors stand out in accuracy while single-stage YOLO models are more suitable for real-time applications. The study highlights that selecting the optimal model is mostly dependent on factors, such as computational resources, allowable delay in detection, and dataset complexity.