Paid gig: negotiable
Get Facebook AI's Detectron – https://research.fb.com/facebook-open-sources-detectron/ – running on mobile HDM device – ODG R7 (using webcamtexture or other method). However, since Detectron requires GPU and ODG R7 has none, need to do processing on server (in my case a windows machine).
What does good look like?
The user running the app on the ODG R7 can see semantic segmentation running in real time on their glasses with highest FPS possible (ie 30FPS-60FPS would be great)
What i think this requires (but you may have a better suggestion):
1) light client (built from Unity3D) runs on ODG R7 which runs android
2) captures frames from client and passes/saves them to a folder/the Unity server via LAN wifi (running on Windows10 Machine with Ti 1070)
3) server passes frames to another service on the windows machine which uses Detectron for semantic segmentation and passes detected objects masks back to client
Any questions, please PM me. Thanks!