The second-hand car market in Romania is continually growing, registering over 660.000 transactions in 2023, according to data from Autovit. However, the market expansion is shadowed by challenges, particularly the inflated prices of used cars. In response to these market dynamics, this thesis introduces a web application under the brand RoCar, tailored for the Romanian automotive market, designed to predict vehicle prices.
RoCar utilizes a model specialized in Romania's economy and pricing, trained on data we have independently scraped. The application differentiates itself from other similar options by using not only structured data such as the year of production, manufacturer, model, and various options but also information extracted from images and descriptions, leveraging a multimodal architecture for a more accurate prediction.
The model is accessible for inference both through a web interface and via an API, thus facilitating the usage by both individual users and businesses wishing to integrate the model into their workflows.
The full application development process and implementation details can be found in the thesis paper here.
