Diagnose

The Spectacular AI SDK includes a self-diagnostic tool to quickly diagnose any recording made with it for potential issues. These include common hardware and configuration issues such as:

  • Data gaps

  • Wrong IMU units (e.g., DPS instead of rad/s)

  • Intolerable mechanical noise levels

  • Incorrect timestamps

To use the tool, simply install it from PyPi and point it at your recording done with the SDK:

pip install "spectacularAI[cli]"
sai-cli diagnose session_2025-07-11_14-46-48/ report.html

To see more detailed options, you can run:

sai-cli diagnose --help

An example report looks like this; see the full report here:

In addition, the data from the SDK can be visualized and diagnosed using the following open source tools:

  • visualize.py script for interactive IMU signal figures.

  • sync-imu-video.py script for assessing IMU-camera time sync. Note that this tool only works in cases where the input data has sufficiently good visual quality such as relatively short handheld tests with versatile (and not too fast) motion, in a feature-rich indoor environment.