Release notes for Scout 2.0
This new version of Scout brings faster and more streamlined task creation and image processing. New machine learning with improved animal detection capabilities, hot keys shortcuts, new search filters, and much more provide extensive improvements to Scout’s functionalities and user experience.
Software DOI for Scout
To reference Scout in a publication, use the following Scout DOI:
About Scout
Wild Me’s Scout is designed to support processing aerial survey data, and is being used to support annotators and lab leads working on the Kavango–Zambezi Transfrontier Conservation Area (KAZA TFCA) aerial survey. Find out more with our Scout Documentation.
New Feature Development
- ScoutBot machine learning provides greater accuracy in detecting animals in images.
- New hot keys shortcuts for labels increase image processing speed and ease.
- A new search filter enables an image search using any part of a filename.
- The Whole image classifier (WIC) score can now be used to filter images.
- Tags for filtering tasks can be added to tasks during their creation.
- Image views are improved, maximizing screen space for images.
- Annotation labels can be created manually as well as generated by the system.
- Editing, viewing, and toggling annotation labels is easier.
- Annotations can be manipulated more easily.
- Images can be ordered either sequentially or randomly.
- Images can be previewed and deleted more easily from the task list view.
Bug Fixes
- The icon display is fixed.
- The annotation interface now processes more than 20 duplicate images both manually and by the ML classifier without issue.
- Annotations and labels now remain fully visible in the Ground Truth reference task view.
- All recorded annotations for an image now appear correctly on the Ground Truth screen.
Known Issues
- When using the Firefox browser, gallery access and task submission are not possible.
- A page refresh is required to see the annotate option in the task table when redirected after creating a task.
- When using a cloud server, an image or annotation export does not get completed.
- Sequenced images are presented in the same order during annotation and ground truth, regardless of camera side sequencing.
- Adding a new tag to a task may require a page refresh to save the tag successfully.