Road infrastructure degradation poses serious threats to public safety and traffic efficiency, especially in emerging nations like India. Road quality improvement, accident risk reduction, and prompt repairs all depend on pothole identification and precise dimension assessment. The suggested system offers a scalable approach to road condition monitoring by showcasing the possibility of real-time deployment in situations with limited resources. In this study, we present an optimized approach to pothole detection and measurement using a fine-tuned You Only Look Once, version 8 (YOLOv8) deep learning model. Another objective is to determine depth of the pothole using monocular vision method. The monocular vision approach provides cost-effective depth estimation, enabling accurate width, length, and depth measurements of detected potholes. According to the results, potholes can be detected with high precision and recall rates, and the dimensions for repair prioritization may be accurately estimated. Our approach has significantly improved performance when compared to existing machine learning methods.