Quick start guide#


To install the TensorBoardPlugin3D python package:

$ pip install tensorboard-plugin-3d


You will need to have TensorBoard installed in order to use this plugin. Information on getting started is available here.

Standalone Appplication#

Select the log directory that contains your 3D image data and run:

$ tensorboard --logdir ~/path/to/logdir

If you’d like to test the plugin with some sample data you can download the cached spleen segmentation data, extract the directory to the location you’d like, and then point to the data:

$ tensorboard --logdir ~/path/to/extraction/location/demo_spleen_data

Open TensorBoard and select the TensorBoard 3D tab.


We have a few notebooks available that allow you to quickly see the plugin in action:

  • UNet Segmentation 3D - Use randomly generated data to quickly train a dataset and view the output.

    Open in Colab
  • Cached Spleen Segmentation 3D - Used cached data from the spleen_segmentation_3d notebook and view the output in TensorBoard.

    Open in Colab
  • Cached Brats Segmentation 3D - Use cached data from the brats_segmentation_3d notebook and view the output in TensorBoard.

    Open in colab