Learning infrastructure setup
We are getting our hands on huge amounts of data while producing video. Efficient ways to parse the data and build a database suitable for training detection and behavioral analytics algorithms will be a fundamental part of future development.
Develop mechanisms to automatically store metadata out of the live footage and the camera operation from every Seervision robot camera (labeling in machine learning). Format footage and metadata for use in (un)supervised machine learning tasks. Develop software bridges between footage metadata and commercial video processing software.
Take full control of the machine learning tasks and focus solely on R&D in close collaboration with ETH. Control over team hiring decisions.
2+ year in machine learning, preferably with neural networks. Experience with video processing software (e.g. Adobe Premiere).