Madhu Korada
Madhu Korada
No, I wasn't able to use. With few libraries such as MACE, MNN GPU was accessible. But 306 seems to be very weak, it's not impactful for inferencing neural networks...
That was my intention as well. Wasted 3-4 weeks exploring the workarounds. 1. In Linux it's not possible to access GPU, have to transition to Android completely to achieve that....
for video??
I had to resolve all the issues related TF and it worked for me.
Yes, Any alternative in tensorflow 2 for ```_linear``` function ?
Simple. ROLO gives you more accuracy. If your application doesn't demand accuracy, you can use yolo and a simple tracker itself.