LiteRT segmentation fault when loading delegate

Hi all. I’m currently getting to know the wonderful world of tensorflow, but ran into a little problem. I want to use the load_delegate function, but when loading the delegate, the software crashes with a segmentation fault.

Hardware: Khadas VIM3
OS: Ubuntu 24.04 (Linux Khadas 5.15.137 #1.7.4 SMP PREEMPT Wed Apr 23 10:46:01 CST 2025 aarch64 aarch64 aarch64 GNU/Linux)
ARM NN: v24.11
Python version: 3.10.17
Python packages:

  • ai-edge-litert==1.3.0
  • numpy==2.2.6
  • pillow==11.2.1

The segmentation fault occurs when running the mobilenet_v1 example from this project: GitHub - sravansenthiln1/armnn_tflite: TFLite examples leveraging the capabilities of the ArmNN TFLite delegate
That project uses the old tflite-runtime package, which also segfaults, but I migrated to LiteRT (ai-edge-litert) but the segfault still exists.

I already tried with multiple python versions, multiple ARM NN delegate version, etc. Hopefully someone can guide me in the use of the correct version and get me past this segmetation fault.