Hi,
I want to run object classification algorithm on 4 live streams (RTSP) in parallel on RK3568 SOM. Since there is no direct support for rockchip MPUs from Edge Impulse, so can i run the model that is created for Raspberry Pi on RK3568? Or will I have to go for some different approach.
Hi @AhsanMoten
Speaking with @ivan and @jbuckEI you should have no issue in running without NPU and it will work as a normal Linux deployment.
If you need some investigation on NPU please speak with our sales and enterprise team.
Contact Sales
Best
Eoin
Thanks @Eoin, I spoke with the sales team on running on NPU , they said they have done it before and its possible with a lot of custom work. Can you please guide me with the process because I want to run inference on 4 parallel camera streams and that will need a lot of processing power.
Is there any way I can get Pytorch or ONNX model format form Edge Impulse training?