In 2018, the United States Department of Transportation Federal Highway Administration reported that Americans traverse over 1,000 miles a month on average. Maintenance of pavement marking conditions is crucial to creating a safe driving environment. Traditionally, pavement markings are assessed periodically through manual road inspectors. This method is time consuming and ineffective in capturing the deterioration of the pavement markings throughout their service life. The incorporation of Unmanned Aerial Vehicles (UAVs) and machine learning techniques provide promising ground to detect pavement marking defects. The study proposes a system based on Deep Convolutional Neural Network (DCNN) that can collect, store, and process data to predict pavement marking conditions throughout their service life. The developed pavement marking detection tool has four stages. The Data Collection stage consists of obtaining images using different methods. In the Classification stage, a method for assessing the defects from images is created based on current methods by Departments of Transportation (DOTs) and the Manual on Uniform Traffic Control Devices (MUTCD). The third phase? the Model Development and Data Processing? and the Model Testing phase are to train and test the model using a multi-level classification program and complex algorithms to process the data collected and output a result. The system was implemented, and the preliminary results show that the model can identify and classify the pavement marking defects. The developed system will help transportation authorities identify and forecast future deteriorating rates and intervention timing.