open-mmlab / mmtracking

OpenMMLab Video Perception Toolbox. It supports Video Object Detection (VID), Multiple Object Tracking (MOT), Single Object Tracking (SOT), Video Instance Segmentation (VIS) with a unified framework.
https://mmtracking.readthedocs.io/en/latest/
Apache License 2.0
3.56k stars 598 forks source link

Failing to build wheels of mmtrack #926

Open tericalpha opened 11 months ago

tericalpha commented 11 months ago

When am trying to install mmtrack , building wheels is getting failed.

RoyAn2386 commented 10 months ago

Hi Everyone, Who's trying to install mmtracking.

It's really complicated when we install it. I stuck with it for 1 week and then found out one easier way to install it. I'll share with you and hope it helps you guys.

Mmtracking

  1. Pull this image from Docker Hub:

    docker pull uestc417/tracking

  2. Make container from the above image file:

    docker run -it --name an_gpu_2_mmtracking --shm-size=16GB --ipc=host -v /home/user/storage/d_ssd/an:/an --gpus '"device=2"' uestc417/tracking:latest

  3. Attach into the above container, git mmtracking by this command:

    **git clone https://github.com/open-mmlab/mmtracking.git**

  4. Move into mmtracking folder and verification:

    cd mmtracking

    install requirements:

    pip install -r requirements/build.txt pip install -v -e . # or "python [setup.py](http://setup.py/) develop"

    python demo/demo_mot_vis.py configs/mot/deepsort/sort_faster-rcnn_fpn_4e_mot17-private.py --input demo/demo.mp4 --output mot.mp4

    If use MOTChallenge evaluation plz install extra dependencies follow:

    pip install git+https://github.com/JonathonLuiten/TrackEval.git

    Got some errors?

    1. Upgrade pip conmand: /root/anaconda3/bin/python -m pip install --upgrade pip
    2. Upgrade numpy: pip install --upgrade numpy
    3. Upgrade pandas: pip install --upgrade pandas

    If the error shown about “QT”, “xcb” problems, try some solutions follows:

    1. Use opencv headless instead of opencv
    pip uninstall opencv-python
    pip install opencv-python-headless
    1. Install this dependencies:

      apt-get install '^libxcb.*-dev' libx11-xcb-dev libglu1-mesa-dev libxrender-dev libxi-dev libxkbcommon-dev libxkbcommon-x11-dev

    2. Turn off display function:

      export DISPLAY=:0

      export QT_QPA_PLATFORM=offscreen

  5. Done!