ImportError: libhailort.so.4.20.0: cannot open shared object file: No such file or directory

I try to run the example streaming by python from github code example. I follow tutorial to the last step : running the yolox_stream_inference.py then I got an error like:
./yolox_stream_inference.py
Traceback (most recent call last):
File “/home/aaeon/Hailo-Application-Code-Examples/runtime/python/streaming/./yolox_stream_inference.py”, line 8, in
from hailo_platform import (HEF, Device, VDevice, HailoStreamInterface, InferVStreams, ConfigureParams,
File “/home/aaeon/.local/lib/python3.10/site-packages/hailo_platform/init.py”, line 17, in
import hailo_platform.pyhailort._pyhailort as _pyhailort
ImportError: libhailort.so.4.20.0: cannot open shared object file: No such file or directory

I go to .local/lib/python3.10/site-packages/hailo_platform/pyhailort and there are some file (pycache, control_object.py, ethernet_utils.py, hailo_control_protocol.py, hw_object.py, i2c_slaves.py, init.py, power_measurement.py, _pyhailort.cpython-310-x86_64-linux-gnu.so, pyhailort.py).

Hi @hung.dd,

Welcome to the Hailo Community!

Could you please check your version numbers for both HailoRT and TAPPAS Core? It seems there might be a version mismatch between the two components.

You can check the versions using these commands:

For HailoRT:

hailortcli --version

For TAPPAS Core:

pkg-config --modversion hailo-tappas-core

Please share the results so we can help resolve any compatibility issues.

I followed this guidance (https://github.com/hailo-ai/Hailo-Application-Code-Examples), and it is not required that hailortcli and tapas be installed. I thought tappas were just examples. I will install hailocli and tappas and try again. I drew a figure to represent my thoughts about the hailo workflow. I don’t know whether it is correct. Can you check it for me? Thank you!

Vào CN, 12 thg 1, 2025 vào lúc 19:56 omria via Hailo Community <notifications@hailo.discoursemail.com> đã viết:

Hi @hung.dd
The suggested hailortcli --version is just a way to quickly check the HailoRT library version installed in your system.
While TAPPAS is not mandatory, the hailortcli tool should be available, as it is installed together with the HailoRT library.

Regarding your scheme is not 100% correct. It is more like this:
The Python APIs (bindings)-> HailoRT → HailoRT PCIe Driver.

The error you are seeing may be due to a missing HailoRT library, or a misalignment in the PyHailoRT/HailORT versions. How did you perform the installation?

If you used the .whl file, please be aware that it just installs the Python bindings, and not the HailoRT library.
Depending on your system, the HailoRT library can be installed in different ways (.deb file, sources, …). Please refer to the HailoRT User Guide for the detailed procedure.

pi@raspberrypi:~/tappas/apps/h8/gstreamer/raspberrypi/detection $ ./detection.sh
WARN: HailoRT version is 4.19, expected to be 4.20 (version was extraced using the following command: ‘hailortcli fw-control identify’)

Hello,
I do every thing from start, i install hailort(deb) → hailo PCIE(deb) → create a venv → hailort(.whl)-> model zoo ->dependency like numpy(1.x version) ->opencv → run example code “yolox_stream_inference.py” get an error (need tensoflow)-> install tensorflow ->run example code again and get this error

Please help me!!! thank you.

Vào Th 2, 13 thg 1, 2025 vào lúc 20:28 pierrem via Hailo Community <notifications@hailo.discoursemail.com> đã viết: