Skip to content

Home

This is MBARI documentation for creating benchmark datasets and evaluating models using HOTA.

A majority of the code in this repository is the TrackEval GitHub repository. I added a bit of code here and there and set up the directory structure to run the HOTA metric for MBARI's benchmark dataset and renamed it benchmark-eval.

Overview

This documentation provides instructions on how to:

  • Annotate videos using RectLabel
  • Obtain ground truth files and model detection files
  • Calculate HOTA scores

Getting Started

  • Check Installation on setting up your environment.
  • Look into Usage on how to navigate the repository
  • See Tips for an explanation of the directory structure and example files.

We hope this is helpful. If you have any questions you can reach the team through GitHub.