I seriously love all of this! I've been playing with this all afternoon. I trained a dataset, I connected a mems mic to an esp32 and now I can control neopixels with my voice. This is serious stuff
Hi, thank you for this video, I remark that there was a c library which not available right now, so is it possible to use a c++ library in my c code or can I add the firmware model to my code?
You can compile the inference code to a library, then link from a C application. See github.com/edgeimpulse/example-standalone-inferencing-c for an example.
Hi, I am using NUCLEO-F446RE, I think the only way I can connect the board to EI is through data forwarder, but the instructions in the documentation is a bit confusing to me. It says load any application and run $ edge-impulse-data-forwarder I am using Windows, does this means I need to have a virtual machine to run that on Linux?
Is this really continuous though? The 2 second intervals really suck. I deployed my model to Arduino library, where there is a regular sketch and a continuous sketch. The regular one with the 2 second interval is quite accurate, but the continuous one basically doesn't work. Why is this?
Hey, the continuous one is the exact same model, but some smoothing applied. You can just get rid of the the smooth calls and print raw predictions to get the exact same output. If that doesn't fix it please create a post on forum.edgeimpulse.com - easier to help from there!
Yes on macOS and Linux via docs.edgeimpulse.com/docs/edge-impulse-for-linux, or alternatively via WebAssembly on all other OS: docs.edgeimpulse.com/docs/through-webassembly
I seriously love all of this! I've been playing with this all afternoon. I trained a dataset, I connected a mems mic to an esp32 and now I can control neopixels with my voice. This is serious stuff
Guess it's ahead of where I'm at currently. Thanks. Maybe after I do some more research first.
Sign me up! It’s mind-blowing demo. I’ll definitely using it in my product for cattle behavior recognition.
Andri Yadi Thanks! Just head to studio.edgeimpulse.com and start building!
Fantastic work!
hihi where can we get this mircocontroaller
Fabulous! Great work!!
Hi, great Demo. Thank you. Just curious, what accelerometer is being used for this demo? Thanks
This is amazing!
Why the frequency domain graphic does´not appear for me ? Someone else with this problem ? Only the power spectral plot appears. I've tried everything
Hi Marco, the graphs were tweaked slightly and the freq domain graph was dropped.
Hi, thank you for this video, I remark that there was a c library which not available right now, so is it possible to use a c++ library in my c code or can I add the firmware model to my code?
You can compile the inference code to a library, then link from a C application. See github.com/edgeimpulse/example-standalone-inferencing-c for an example.
Hi, I am using NUCLEO-F446RE, I think the only way I can connect the board to EI is through data forwarder, but the instructions in the documentation is a bit confusing to me. It says load any application and run $ edge-impulse-data-forwarder
I am using Windows, does this means I need to have a virtual machine to run that on Linux?
No, just open a command prompt and run `edge-impulse-data-forwarder` - works on Windows too.
Is this really continuous though? The 2 second intervals really suck. I deployed my model to Arduino library, where there is a regular sketch and a continuous sketch. The regular one with the 2 second interval is quite accurate, but the continuous one basically doesn't work. Why is this?
Hey, the continuous one is the exact same model, but some smoothing applied. You can just get rid of the the smooth calls and print raw predictions to get the exact same output. If that doesn't fix it please create a post on forum.edgeimpulse.com - easier to help from there!
@@janjongboom7561 ah got it, I'll give it a shot, thanks for the response!
Can I used PC to run this model ?
Yes on macOS and Linux via docs.edgeimpulse.com/docs/edge-impulse-for-linux, or alternatively via WebAssembly on all other OS: docs.edgeimpulse.com/docs/through-webassembly
That's dope af