Using SSD and YOLO to efficiently detect potholes and speed bumps is the main goal of this project, which aims to improve road safety and construction quality. Hazard priority is emphasized and the limitations of the manual method are addressed. YOLO and SSD architectures are used, trained, and optimized on a variety of datasets with different road appearances and conditions. To improve model capabilities, data augmentation approaches are used. The study explores techniques for processing data quickly through optimization. Accuracy, recall, and efficacy are measured by gathering large amounts of data, training models, and assessing results on different test sets. Extensive analysis between YOLOv4 and SSD-MobileNet demonstrates clear differences in performance. YOLOv4 achieves an overall improved object detection proficiency of 85.48, a superior mAP@0.5 score. By comparison, SSD-MobileNet demonstrates unique metric strengths in precision (0.42), recall (0.81), and F1 score (0.82). In particular, SSD-MobileNet significantly outperforms YOLOv4 (52.51 ms) in inference time (7 ms), demonstrating its effectiveness in real-time applications. The study suggests using SSD-MobileNet because of its excellent F1 score, repeatability, and precision for real-time applications. Because of its quick inference time, this model is a great fit for high-stakes situations where accurate and timely object identification is necessary, such as security systems, drones, and autonomous cars. It also balances performance measurements with practicality. Novelty: By utilizing cutting-edge computer vision techniques, this study creatively combines SSD and YOLO to produce a safer and more effective method of infrastructure management and road danger monitoring.