Hello @LinuxMan,
Thanks for all this information.
NXP confirmed to me:
- neutron-converter is not required for Linux 6.6
- It is only mandatory starting from Linux 6.12
I was not aware of this.
Anyway, from your output logs I see that you have a Verdin iMX95 SoM and NOT a Verdin iMX95 EVK. Could you please tell me how you built the BSP with the machine learning libraries for this SoM? I need to know your steps to know what is going wrong.
The reason I am asking this is also that I could not reproduce the error that you are seeing. I built a BSP reference multimedia image for Verdin iMX95 with the tensorflow libraries and tried to run the example mobilenet model without converting it with Neutron converter. Here is my output:
root@verdin-imx95-12593795:~# cat /etc/os-release
ID=tdx-xwayland
NAME="TDX Wayland with XWayland"
VERSION="7.4.0-devel-20251203102051+build.0 (scarthgap)"
VERSION_ID=7.4.0-devel-20251203102051-build.0
VERSION_CODENAME="scarthgap"
PRETTY_NAME="TDX Wayland with XWayland 7.4.0-devel-20251203102051+build.0 (scarthgap)"
CPE_NAME="cpe:/o:openembedded:tdx-xwayland:7.4.0-devel-20251203102051+build.0"
root@verdin-imx95-12593795:~# uname -a
Linux verdin-imx95-12593795 6.6.101-7.4.0-devel #1 SMP PREEMPT Thu Sep 25 07:49:28 UTC 2025 aarch64 GNU/Linux
root@verdin-imx95-12593795:~#
root@verdin-imx95-12593795:~# cd /usr/bin/tensorflow-lite-2.16.2/examples/
root@verdin-imx95-12593795:/usr/bin/tensorflow-lite-2.16.2/examples# ./label_image -m mobilenet_v1_1.0_224_quant.tflite -i grace_hopper.bmp -l labels.txt --external_delegate_path=/usr/lib/libneutron_delegate.so
INFO: Loaded model mobilenet_v1_1.0_224_quant.tflite
INFO: resolved reporter
INFO: EXTERNAL d[ 380.166724] remoteproc remoteproc0: powering up neutron-rproc
[ 380.174777] remoteproc remoteproc0: Booting fw image NeutronFirmware.elf, size 42612
[ 380.182920] remoteproc remoteproc0: remote processor neutron-rproc is now up
INFO: NeutronDelegate delegate: 29 nodes delegated out of 31 nodes with 1 partitions.
INFO: Applied EXTERNAL delegate.
INFO: Created TensorFlow Lite XNNPACK delegate for CPU.
INFO: invoked
INFO: average time: 1.675 ms
INFO: 0.729412: 653 military uniform
INFO: 0.164706: 907 Windsor tie
INFO: 0.0196078: 458 bow tie
INFO: 0.00784314: 835 suit
INFO: 0.00784314: 466 bulletproof vest
root@verdin-imx95-12593795:/usr/bin/tensorflow-lite-2.16.2/examples# [ 457.153052] remoteproc remoteproc0: stopped remote processor neutron-rproc
As you can see, it runs correctly here on my side. So I need to see your steps for how you built the image to help you further.