I recently created a project deployment library for object recognition.
Arduino Studio, version 2.2.1
My camera module is an ESP32-WRover, but it has been configured for development, therefore it is ESP32 Dev Module
I get this
tensorflow-related error when i try to compile the ESP32 Cam example that came with my deployment.
edge-impulse-sdk/tensorflow/lite/micro/kernels/softmax.cpp:301:14: error: return-statement with a value, in function returning 'void' [-fpermissive]
and my IDE hangs. the only way to restart it is by doing a
killall -9 arduino-ide
Edge Impulse deployment work with an
ESP32 Dev Module?
what is also interesting is when i try to verify the
static buffer example, i get the same error listed above.
Please advise. Your help will be greatly appreciated.
Thanks in advance,
please disregard this issue. i resolved it by deleting the
libraries directory and reinstalling my dependencies.
thanks all the same.
I ran into this as well, and I had to make source code changes to fix the compiler errors. For example, the example mentioned above is a void function that returns a value, which will result in a compiler error. I haven’t found another workaround, since the source code with the errors is the code exported into the zip file when deploying as Arduino.
I wonder if there’s a way to submit code fixes?
In my case, to repro the issue:
- Deploy any model as Arduino
- In Arduino IDE, Sketch->IncludeLibrary->AddZipLibrary
- Select Xiao ESP32-C3
- Build Static Buffer example
I had to make small changes in the below files:
hi @MicroMasjien ,
thank you for your response.
i discovered why my examples will not compile. i was using
1.0.6. this is the last version before
2.0.1. from version
2.0.1 onwards, the examples compile and deploy to my
ESP32-WRover, but the camera will not initialize.
Interesting, I’m currently on “Arduino Alpha 3 Release v3.0.0”
I also get the error but only for the unoptimized (float32) version of the Arduino library, the quantized (int8) version compiles smoothly.