VOXL Deep Learning 30FPS Using TensorFlow Lite on GPU for Linux

TensorFlow Lite port using GPU Delegate for Mobilenet v1 on ModalAI VOXL

VOXL is an open embedded Linux system. TensorFlow Lite was modified to support OpenCL GPU processing on Linux. Officially TensorFlow Lite only supports the GPU Delegate on Android.

MobileNet v1 was used as the full workflow for training and inference is readily available. DeepLab v3 runs in this framework even much faster!

==========
RESOURCES
==========

==============
ABOUT MODALAI
==============
ModalAIⓇ accelerates development of smaller, smarter and safer robots and drones with Blue UAS Framework autopilots built in the U.S.A.
Based in San Diego, California, ModalAI spun out of Qualcomm® in 2018 and leverages over five years of prior research and development in the drone and robotic markets.