Skip to content

Latest commit

 

History

History
100 lines (86 loc) · 4.88 KB

File metadata and controls

100 lines (86 loc) · 4.88 KB

Installation Instructions

Installation from precompiled packages [latest version not available]

  • Download python 3.6>= using anaconda or miniconda if you prefer a lighter installation.
  • Open a Terminal in OSX or Linux. In Windows you need to open the Anaconda Prompt.
  • [Optional] I would recommend to create and activate an enviroment as:
conda create -n tierpsy

conda activate tierpsy #[Windows]
source activate tierpsy #[OSX or Linux]
  • Finally, donwload the package from conda-forge
conda install -c conda-forge tierpsy numpy=1.16.3 pytables=3.5.1 opencv=3.4.8 pandas=0.24.2 #[Windows]
conda install tierpsy -c conda-forge tierpsy python=3.6 opencv=3.4.2 'pandas<1.0' #[OSX]
  • After you can start tierpsy tracker by typing:
tierpsy_gui

Do not forget to activate the enviroment every time you start a new terminal session.

On OSX the first time tierpsy_gui is intialized it will create a file in the Desktop called tierpsy_gui.command. By double-cliking on this file tierpsy can be started without having to open a terminal.

Troubleshooting

  • When installing from conda, you may get an error while the package conda-forge::cloudpickle... is installed, stating that python3.6 couldn't be found. In this case, updating conda has been known to solve the issue. Alternatively, make sure to first install python 3.6, and then tierpsy, by executing:
conda install -c conda-forge python=3.6
conda install -c conda-forge tierpsy
  • It seems that there might be some problems with some opencv versions available through conda. If you have problems reading video files or encounter error related with import cv2, then you can try to install opencv using pip as:
pip install opencv-python-headless

or specify the opencv version via:

conda install opencv=3.4.2 #[tested on MacOS]
  • In Windows, the default anaconda channel does not have a valid ffmpeg version. Activate the tierpsy enviroment and use the conda-forge channel instead as:
conda install -c conda-forge ffmpeg

Installation from source [for development]

MacOS

  • Open a Terminal prompt and type:
git clone https://github.com/Tierpsy/tierpsy-tracker
cd tierpsy-tracker
conda env create -f tierpsy_mac.yml #[MacOS]
conda activate tierpsy
pip install -e .
tierpsy_gui

Tested on Mojave

Windows 10

  • Clone the repository in a folder named tierpsy-tracker. You can do this either:
    • via a git prompt
    • with your browser: Clone or Download -> Download ZIP, then unzip and rename the folder
    • with your browser: Clone or Download -> Open in Desktop, then continue with GitHub Desktop
  • Open the Anaconda prompt, move to the tierpsy-tracker folder using the cd command appropriately, and type:
conda env create -f tierpsy_windows.yml #[Windows 10]
conda activate tierpsy
pip install -e .
tierpsy_gui

Troubleshooting

  • Make sure you have an up-to-date version of conda. To update conda, conda update -n base -c defaults conda. We tested on conda 4.8.2.
  • Try the alternative command conda env create -f tierpsy_windows_conda4_5_11.yml
  • pip install -e . has been known to fail with an error stating that command 'cl.exe' failed: No such file or directory. See the Known Issues for a solution.

Linux

  • Open a shell and type:
git clone https://github.com/Tierpsy/tierpsy-tracker
cd tierpsy-tracker
conda create -n tierpsy #[optional]
conda activate tierpsy #[optional]
conda install --file requirements.txt
pip install -e .
tierpsy_gui

Batch processing from the command line.

The script TierpsyTrackerConsole.py was deprecated in favour of using the command tierpsy_process. Type tierpsy_process -h for help.

Tests

After installing you can run the testing scripts using the command tierpsy_tests on the terminal. Type tierpsy_tests -h for help. Although the script supports running multiple tests consecutively, I would recommend to run one test at the time since there is not currently a way to summarise the results of several tests.