SK-AM62A-LP: Looking into porting existing application which runs on TDA4x architecture to this board

Hi Harmander, Thank you for choosing AM62A as a dev platform. There are a couple of subtle differences between 62A  applications and TDA4. As an example in 62A the capture drivers are on ARM and therefore there are some implications to your dev flow.

If your example is a single camera application pipeline, I would reccomend starting of here in the links below. There are example dataflows put together for a typical 1Mp USB camera or a 2MP auto  camera application using Gstreamer. The applications show how to run multiple models simultaneously and take in multiple inputs. 

https://software-dl.ti.com/jacinto7/esd/processor-sdk-linux-edgeai/AM62AX/09_00_00/exports/edgeai_docs/common/sdk_overview.html

https://software-dl.ti.com/jacinto7/esd/processor-sdk-linux-edgeai/AM62AX/09_00_00/exports/edgeai_docs/common/edgeai_dataflows.html

If your application is Deep Learning Heavy,  a good place to start is by first profiling your networks. Please do start with the above data flows and we will also connect with you offline

thanks

Tarkesh