The global pandemic has been a forcing function of everything digital and virtual, whether in classrooms, work environment, shopping, or social life. All of this change, while being very disruptive, is also the beginning of future opportunities.
Global Venture Catalyst is a new initiative focused on creating connections and empowering teams of technology-focused undergraduates, MBA students, and experienced mentors to connect and pursue new ventures or corporate opportunities.
The program is developed by Prof. Ikhlaq Sidhu at UC Berkeley based on the concepts of Innovation Engineering.
Over the past two weekends, I participated in the inaugural Global Venture Catalyst Project Sprint. As mentor, I collaborated with an MBA student based in Prague and three graduate students based in Melbourne, Jakarta and New Dehli respectively. We worked on identifying and analyzing opportunities for Dassault Systèmes to provide the training data (rendered images) required for teaching AI/ML based Computer Vision (CV) to ‘see’ things.
The team was motivated and given the limited time delivered some real value:
- The students received a crash course on CV including state of the art model training approaches using synthetic data.
- 3DEXCITE received validation of the opportunity through conversations with potential CV customers: specifically CV robot camera providers focused on retail use cases (Cortixica and Zebra). Currently their robots drive around retail stores taking photos of products on shelves, which they use as training data to learn what each product looks like. They stated that provision of rendered image training data would eliminate at least 50% of their CV model training costs.