People say that you need to build your house three times to fulfill your dreams.
Hegias, a start-up based in Lugano and Zurich has developed a platform with which you can visualize, explore and modify your house in the virtual reality world already from the initial planning phase.
This innovative software allows you to immerse yourself in a virtual reality reproduction of your 3D scenes and models.
In this product, the classification of 3D objects plays a very important role and for this reason, Hegias asked me, as a machine learning expert, to improve the classification system and reduce human intervention to a minimum to speed up the process but also to reduce the costs.
A demanding task: how can a computer classify an object as a chair, a shelf or as part of the architecture? A machine learning algorithm, thanks to many examples, can learn and improve and it eventually can understand how to classify objects that it has never seen before.
As a practical example, if you show to the algorithm this chair here, this one here and this one here, it learns to look at the corresponding point in the 3D shape that this one here is also a chair.
After a complete review of the state of the art of machine learning techniques for 3D objects classification and the development of a new innovative approach, I was able to create a machine learning prototype that achieved a classification accuracy of 92%.
Thanks to the outstanding supervision of Professor Cesare Alippi and my hard work, this project allowed me to finish my Master Degree in Artificial Intelligence (USI) with exceptional results winning the Swiss Engineering Ticino Prize for the best Master thesis in Informatics, but more importantly, to see my work in use by a startup which is revolutionizing the architecture industry by allowing you to build your dream house without making expensive mistakes.