After three days of brainstorming, coding and networking, Sreejit Dutta together with his team members Katarina Golubovic, Amélie Guého and Rupesh Kumar won the BMW-AWS "AI4RepairConcept" hackathon, which took place at the BMW Group Digital Campus Munich from November 4 to 6.

According to Dutta, their team was the only group that implemented a machine learning (ML) model for the whole challenge.

“We actually made one mathematical model, and one machine learning model. The model we implemented is called PointNet and it is used for 3D segmentation and classification,” Dutta says.

The hackathon, organized by BMW Group in collaboration with Amazon Web Services, invited developers and engineers to participate in a challenge aimed at developing AI-based techniques for repair concepts based on 3D geometric models of the company's cars.

“From figuring out 3D Geometric data for cars and coding Machine Learning models, to implementing them on AWS, it was a fascinating experience, one I would remember forever. All that hardwork and effort lead us to being the winners at the hackathon,” Dutta added.

Sreejit Dutta has been working as a research assistant at the Chair of Remote Sensing Methodology since September 2020. He is doing his Masters in Data Engineering and Analytics, working with classical and quantum mechanical ML methods for time series analysis of satellite data for crop detection, part of the EuroCrops project.