Develop a mono-depth network using the Intel Realsense stereo camera

External Member

Nariman Habili


Abstract: Use the Intel Realsense stereo camera to train a mono-depth network, where the output pixel values of the depth image indicate distance in meters. Or retrain an existing network, such as mega-depth so that the output pixel values are distances in metres and not relative distances. Students need to collect data using the camera. A GUI that shows the Realsense camera outputs (RGB and/or Depth) in a window and the results in another. Tools required: Python, PyQt, Pytorch, Realsense Python API.

Contact: Nariman Habili


Updated:  1 June 2019/Responsible Officer:  Dean, CECS/Page Contact:  CECS Marketing