Quality control and data curation workflows are vital to ensure data is findable, accessible, interoperable and reusable (FAIR, Wilkinson et al., 2016). All corrections should be made within the original annotation files to ensure data consistency over time. We recommend the following approaches to ensure quality control:
- Annotators should complete small identical ‘training’ image sets where habitat classes are known, to assess competency and benchmark accuracy.
- Quality assurance should be carried out by a senior analyst and involves a randomised review of 10% of annotated images and data within a project. If accuracy is below 95% for all identifications, imagery should be re-annotated.
- All annotators should meet periodically as a group to discuss image classification to ensure that consistency is maintained throughout the project.
We propose a series of simple visual quality control plots to identify outliers and provide examples of these in the annotation guide (globalarchivemanual.github.io/CheckEM/, Fig. 1).