Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Learning based real-time depth estimation using a front mounted stereo camera on a motorcyclele to enable accurate AR anchor generation
Safety systems for motorcycles are lagging several years to that of cars. That is in part
due to the smaller form factor of the motorcycle and the more unstable dynamic system
with roll and pitch being significant axis of freedom.
Safety systems for motorcycles are lagging several years to that of cars. That is in part
due to the smaller form factor of the motorcycle and the more unstable dynamic system
with roll and pitch being significant axis of freedom. Nonetheless, modern advanced
driver assistance systems require accurate information about the world around the vehicle
in 3D and real-time. A stereo camera can be used obtain this information while staying
robust to roll and pitch, being cheap (significantly compared to lidar) and paired with
advanced learning based approach achieve similar performance.
As part of a joint master thesis between the Computer Vision Lab and Aegis Rider AG an
ETH Pioneer Fellowship startup, you will develop a learning based real-time depth
estimation system using a stereo camera mounted on our prototype motorcycle.
**Requirements**: knowlegde in C++ and Python, knowledge on deep-learning frameworks
such as Pytorch or Tensorflow including embedded optimisations via TensorRT.
Safety systems for motorcycles are lagging several years to that of cars. That is in part due to the smaller form factor of the motorcycle and the more unstable dynamic system with roll and pitch being significant axis of freedom. Nonetheless, modern advanced driver assistance systems require accurate information about the world around the vehicle in 3D and real-time. A stereo camera can be used obtain this information while staying robust to roll and pitch, being cheap (significantly compared to lidar) and paired with advanced learning based approach achieve similar performance. As part of a joint master thesis between the Computer Vision Lab and Aegis Rider AG an ETH Pioneer Fellowship startup, you will develop a learning based real-time depth estimation system using a stereo camera mounted on our prototype motorcycle.
**Requirements**: knowlegde in C++ and Python, knowledge on deep-learning frameworks such as Pytorch or Tensorflow including embedded optimisations via TensorRT.