Munich, Bavaria, Germany
I love working at the intersection of cutting-edge technology and real-world problems. As the co-founder of VisualAIs, I help simplify and accelerate the creation of 3D visualizations, transforming how products are displayed online. We seek customers in e-commerce—reach out if you’re interested! Alongside my startup, I’m a PostDoc at TUM, advising a team of PhD students on AI and Computer Vision topics. I also welcome research collaborations with industry and academia. During my PhD at TUM in Computer Vision, I published papers in top conferences, including CVPR and ICCV, and contributed to autonomous driving perception at BMW. My journey in AI began in 2017 during my MSc at TUM and research at Airbus. I was honored with the Deutschlandstipendium (merit-based scholarship) for 2 years and received Outstanding Reviewer awards at ICCV 2023 and ECCV 2024. If you want to explore how 3D visualizations can enhance your business or talk about AI, Computer Vision, or entrepreneurship, let’s connect! > Startup website: https://www.visualais.io > Personal TUM page: https://www.cs.cit.tum.de/camp/members/stefano-gasperini > Google Scholar page: https://scholar.google.com/citations?user=YuWTPaIAAAAJ
We develop AI-based solutions to simplify the creation of 3D renderings, directly from smartphone images! We are supported by UnternehmerTUM, XPLORE, and TUM Venture Labs Software and AI. In 2023, we obtained a grant from the Bavarian Ministry of Economic Affairs, Energy and Technology. Get in touch if you're interested!
We are open to collaborations with research institutes and industry. Reach out if you're interested! - Advising 6 PhD students on various Computer Vision topics (including 3D reconstruction and scene understanding), with PD Dr. Federico Tombari. - Senior member of the Computer Vision Team at the CAMP Chair. - Activities mainly involve advising PhD students, seeking new projects, funding opportunities, and collaborations with industry and other research partners, as well as coordinating a research group about 3D reconstruction. - Recipient of the Outstanding Reviewer Award at ECCV 2024. This award was also obtained at ICCV 2023.
- 6 first-author research papers with BMW affiliation. Papers published at CVPR, ICCV, IJCV, RA-L + ICRA, and 3DV (incl. equal contribution). - 6 patents filed with BMW during this time. - Research on Reliable Scene Understanding for Autonomous Driving with Computer Vision. - PhD done at TUM with research carried out at BMW (i.e., TUM external PhD). - ProMotion program by BMW. Info here: https://www.bmwgroup.jobs/de/en/students/entry-programmes/phd-programme.html - Supervised 5 Master's Thesis students at both BMW and the TUM side. I was their primary contact on both ends and led several thesis projects to publications. - Partly involved in VDA KI-Absicherung: a research project with the German Automakers Association (VDA), featuring a consortium of industry and academic partners working together to enable trust in Artificial Intelligence systems for highly Automated Driving. Project sponsored by the German Federal Ministry for Economic Affairs and Energy (grant number: 19A19005B).
- Scientific paper titled "Signal Clustering with Class-Independent Segmentation" published at IEEE ICASSP 2020, best in signal processing. - Research work on Deep Learning applied to clustering and signal separation by source. - I came up with a novel clustering method based on a Neural Network, trained to cluster the elements directly, without the need for traditional clustering approaches from data analysis, which previous methods exploit to group the inputs. - Thesis project for my Master of Informatics at TUM.
- Worked 15h/w for 1 year on an AI research project that started when I joined at the beginning of April 2018. Done during my Master's. - I was responsible for a software demonstrator and its Artificial Intelligence components: developed a prototype and applied Machine Learning and Deep Learning to classify signals.
- Business line: Wealth and Asset Management. - Contract through Modis until June, then hired directly by Prometeia. Worked full-time from early December until the end of September. - My work first focused on the build lifecycle, therefore I curated the build process for some of the company projects. - I personally developed a program using Node.js and Maven to run on Jenkins for continuous integration. My program fully automatizes and parallelizes the builds (including deploy and release) of complex work-in-progress projects with 85+ “snapshot” dependencies organized in 15+ levels of depth. This allowed to reduce runtime by up to 95% for tasks run multiple times per day (over the handcrafted solution in use before), while automatizing other processes previously done manually overnight. - Afterwards I continued maintaining the projects build processes and I also worked on a vertical application based on the Java ecosystem, with tasks including refactoring, technology updating and customization for new customers.