The idea is that I can validate the results of my HW-accelerator / network. Therefore, I want a C-runtime image that I can build with the HW-accelerator matching stub/skeleton.
The validation should be done the following:
Elastic-ai.creator Python Environment
- Build your model + train
- Generate HW-Accelerator + matching Skeleton
- Generate binfile
Elastic-ai.runtime 4. Generate matching Stub 5. Build C-Runtime
Elastic-ai.cloud 6. Start your Cloud application incl. MQTT Broker (unfortunatly very unspecific)
Elastic-ai.enV5 7. Flash the RP2040 over USB sad noises :( 8. enV5-Connects via MQTT to the broker 9. enV5 Subscribes to "compute-data"-topic to receive data 10. Flash the binfile via HTTP happy :)
Send Data via MQTT 11. Connect with a MQTT Client 12. Subscribe "computed-result"-topic 12. Send data to "compute-data"-topic 13. Wait till results are received via "computed-result" topic 14. Save the results
Elastic-ai.creator Python Environment 15. Check if the results match the software implementation 16. PROFIT