Potholes on roads result in accidents, structural degradation, fuel consumption, and traffic jams, which makes the speedy detection and repair of these road potholes crucial for better road safety and reducing the economic loss. Traditional methods for inspecting roads are time-consuming, labor-intensive, and prone to errors. This paper presents a system for pothole detection based on a fine-tuned YOLO object detection model implemented in Python to make real-time analysis of static images and video streams compatible with either surveillance cameras or vehicular dash cams. Advanced data preprocessing techniques and Google Colab’s GPU infrastructure are used to train the model quickly, in a fast, efficient, and dependable manner. The system is highly accurate and reliable, even in challenging conditions like wet surfaces and complex road environments. Evaluation metrics, including accuracy, recall, and inference time, support its effectiveness in generating actionable knowledge for local authorities, ensuring timely repairs and better transportation facilities while saving maintenance costs and ensuring smoother travels.