search

davidsandberg / facenet

Face recognition using Tensorflow
MIT License
13.85k stars 4.81k forks source link

High Memory Usage #1006

Open gokh4n opened 5 years ago

gokh4n commented 5 years ago

Firstly of all thank you for the Facenet. I added an extra 300 unknown classes to get better accuracy than the CNN model. And then I added my real people. I'm using smoothly now. But I should use the application with Nvidia Jetson Nano. It has 4GB Memory. I use 2.7 gb ram for Facenet recognition and Facenet detection. How can i use memory less and run faster. Thanks for the answers.

harsath commented 5 years ago

How to run the Facenet Model on Jetson Nano Faster with higher FPS on realtime. Or Which model architecture Do I need to use in order to perform high computational and high FPS tasks?