Cannot load built engine resnet50_market1501_aicity156

This problem should not be related to the tracker, it will occur even if the tracker is not used.

There should be no problem with the code, but I reproduced the problem in DS-6.4.

You can use the DS-6.3 version first. There is no problem with this version.

I’m sorry, but I don’t understand your answer.

You manage to reproduce the problem and you suggest as a solution that I use Deepstream 6.3?

I want to use 6.4.

Any news?
I confirm this is working fine for 6.3 but I want to use 6.4.
Here is the Dockerfile for Deepstream 6.3 :

FROM nvcr.io/nvidia/deepstream:6.3-triton-multiarch as base

WORKDIR /opt/nvidia/deepstream/deepstream

RUN apt-get update && apt-get install -y \
    python3-pip libcairo2-dev libcairo2 pkg-config python3-dev protobuf-compiler libyaml-cpp-dev libmpg123-0 libmpeg2-4 libavcodec58 libgirepository1.0-dev \
    # Install necessary packages for building GLib:
    pkg-config python3-dev libffi-dev libmount-dev meson ninja-build wget

RUN python3 -m pip install --upgrade pip && pip3 install \
    Cython packaging meson==1.2.0 pydantic==1.10.7 typing-extensions==4.5.0 python-dotenv pika opencv-python confluent-kafka

# Building GLib 2.79.1 (required for DS 6.4)
WORKDIR /opt/nvidia/deepstream/deepstream/lib
RUN wget https://github.com/GNOME/glib/archive/refs/tags/2.79.1.tar.gz && tar -xzvf 2.79.1.tar.gz
WORKDIR /opt/nvidia/deepstream/deepstream/lib/glib-2.79.1/
RUN rm -r subprojects/gvdb && meson _build && ninja -C _build && ninja -C _build install && rm /usr/lib/x86_64-linux-gnu/gio/modules/libgiolibproxy.so # removing conflicting shared object file
# Set environment variable for new GLib version
ENV GI_TYPELIB_PATH=/usr/local/lib/x86_64-linux-gnu/girepository-1.0
ENV GIO_MODULE_DIR=/usr/lib/x86_64-linux-gnu/gio/modules/

WORKDIR /opt/nvidia/deepstream/deepstream

# Install GStreamer python bindings (whl is not installing all required depedencies)
RUN ./user_deepstream_python_apps_install.sh --version 1.1.8

ENTRYPOINT ["/bin/bash", "-c", "tail -f /dev/null"]
CMD []

I am have exactly the same problem, the fixes on pyds only solved the problem in deepstream-test2 but not in my own custom project, cannot share my code due to privacy issue, but the issue is EXACTLY THE SAME. Had to continue using 6.3 instead for now

1 Like

This may be an issue of DS-6.4. I haven’t confirmed where the problem is yet.

1 Like

@chen8131928
You can open a new topic to discuss the problem. If it is not convenient to share the code, you can also tell me how to use sample code to reproduce the problem.

@junshengy Have you soloved this issue? It does not work on 6.4 but works in 6.3. I tried deepstream 7.0 today, the problem is the same as in 6.4.

It is not solved.
It looks quite a significant bug but apparently, they are not working on it.
In that post, they have all the necessary information to reproduce and debug.

@simon.renuart Thank you for letting me know, it is indeed quite signficiant as no tracker engine can be used for inference. Surprisingly the 7.0 did not fix it still.