Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Growing Scene graphs from Observations
Motivation: Building "dynamic Scene Graphs" that allow to add to the already acquired static information of the
scene data that can be learned from observations based on the users interactions
How: 'Mine' the knowledge from egocentric observations (e.g. aria glasses, HoloLens). Potentially, action
recognition, and tracking, etc. Will be helpful to process these observations. Then build a static scene graph with the information of the environment. Some information is stored explicitly but all the information should be ‘queryable by language’ such that we can answer
Not specified
Not specified
Requirements: experience with a Python deep learning framework, understanding of 3D scene and camera geometry.
Please send us a CV and transcript.
zbauer@ethz.ch blumh@ethz.ch
Requirements: experience with a Python deep learning framework, understanding of 3D scene and camera geometry.