Unity Computer Vision Dataset Visualizer

This Python based tool allows you to visualize datasets created using Unity Computer Vision tools.

Requirements

• Windows 10 or OSX
• Chrome, Firefox, or Safari 14 and newer (Older versions of Safari are not supported)
• Python 3.7 or 3.8. Note that this application is not compatible with Python 3.9.

Installation

We recommend using a virtual enviornment to install and run the app. One way to achieve this is using Conda.

Step 1: Create a virtual environment (skip to step 2 if you are setting up a virtual environment using other methods)

• Install Conda if you do not already have it installed on your computer. We recommend Miniconda.

• Once Conda is installed:

  • On Mac OS, open a new terminal window.
  • On Windows, you will need to open either Anaconda Prompt or Anaconda Powershell Prompt. These can be found in the Start menu.

• Create a virtual environment named dv_env using Conda, and activate it (use 3.7 or 3.8 for <python_version>):

conda create -n dv_env python=<python_version>
conda activate dv_env

Step 2: Install application

Use the following command to install the visualizer.

pip install unity-cv-datasetvisualizer

ℹ️ On Windows, if you get an error about the system not being able to create a process when running the install command, make sure you have the latest version of Conda installed. Then create a new environment with a new name, and try the install command again. If the error persists, try leaving the Conda virtual environment using conda deactivate and running the command outside of any virtual environment.

ℹ️ If you receive an error saying the unity-dataset-visualizer' distribution was not found, try closing your command line window, opening it again and activating the dv_env environment, and running the datasetvisualizer command again.

Running the visualizer

Run the command:

datasetvisualizer

Or if you want to specify a path to a dataset:

datasetvisualizer --data="<path_to_dataset>"

This command may take a few seconds to execute. Once it is done, your browser will automatically open to http://localhost:8501/ and display the application. If that does not happen, open a new browser tab and manually navigate to that address.

Once in the application, you will be prompted to select a dataset folder. Click Change Dataset at the left side of the screen and then select the root folder of your Unity Computer Vision dataset.

Known issues

• The tool cannot open a dataset that has no labeler data (bounding boxes, semantic segmentation, etc.)
• On Windows: a warning appears when launching the app (This can be ignored)
• 3D bounding boxes are not rendered properly when the camera is inside the box.
• Opening a folder that isn't a dataset folder may hang the program (The workaround is: kill the process and start it again)