Enhancing Used Car Price Predictions with Machine Learning-Based Damage Detection

Due to the rising sale of cars, the price of secondhand cars has evolved into a significant concern, which may negatively impact sustainability. Using online platforms to determine the price of used vehicles has become increasingly popular to ease the gap between buyers and sellers. This study contributes to the improvement of the used car industry by enhancing the accuracy of damage detection and price prediction and building trust among buyers and sellers by incorporating a machine learning-based system. It aims to automate the manual work required while buying and selling cars, including damage assessment and price adjustment based on features of the car provided by the user. This paper demonstrates a system that accepts user input in the form of text and images and employs You Only Look Once (YOLO) to detect damages in uploaded images of used cars. The cost of these vehicles is predicted based on relevant features by applying three regression and boosting algorithms to publicly available datasets and selecting the best one.