Question/Issue:
How to solve issue for deployment failed & arduino library won’t upload
Project ID:
193597
Context/Use case:
Hello im new to edge impulse so bare with me 
Im encountering two problems…
I trained an image classifier and in the phase of deployment it succesfully made a zip for an Arduino Library but when I tried to upload it to the Arduino IDE it get stucks: “C:\Users\marco\AppData\Local\Arduino15\packages\arduino\tools\arm-none-eabi-gcc\7-2017q4/bin/arm-none-eabi-g++” -c -w -g3 -nostdlib ". To clarify in the Arduino IDE I have selected the
Arduino Nano 33 BLE board
Because of this, I tried to build the Arduino Nano Ble sense FW option but it displays the next error:
Hi @victorinox,
The error ... section '.text' will not fit in region 'FLASH' means that your model is too big for the microcontroller. You are trying to use AkidaNet, which is specifically designed for the BrainChip Akida chip. Even some of the MobileNet models are too big for the Nano 33 BLE Sense. You will need to choose a smaller model architecture if you wish to use the Nano 33 BLE Sense.
Hello @shawn_edgeimpulse
Thank you very much, does this means that i have to use less images? in the training data (400 images) or in the test data (100 images) also, I ran the impulse directly into my phone and it works.
Hi @victorinox,
The number of training images does not affect the size of the model. The model size is determined by the architecture (and other optimizations, such as quantization and use of EON Compiler). The model you have chosen (AkidaNet) will run on a phone because a phone has a lot more resources (gigabytes of flash storage) as opposed to the Nano 33 BLE Sense (1 MB).
To fit your model onto the Nano 33 BLE Sense, you will need to chose a different model, such as a basic “Classification” model (2-layer CNN). A small MobileNetv2 model may also fit.
Hello! @shawn_edgeimpulse
Thank you very much for your help and the tips!!!
It worked I used the MobileNetv2 96x96 and the EON compiler too.
1 Like
Glad to hear that it worked!
Hello , i had the same error even when i used the mobilenetv2 96x96 it doesn’t work for me. 
