Introduction
Data is more critical than ever in the fast-paced world of computer vision development. But in this case, we don’t mean data in the sense of labeled images or videos for training ML models, but data on how your teams are performing and projects are progressing.
Accurate and up-to-date data or metrics of this kind are essential for informed decision-making at the project management level and knowing how to optimize your workflow and processes for maximum efficiency. However, getting this type of data requires a robust system for labeling management and reporting.
That’s why we continue to work to expand and improve our analytics capabilities. For this release, we’ve introduced a slew of small yet impactful additions to your reporting dashboards. In addition, we’ve introduced a few new SDK functions and other overall performance and annotation improvements. Let’s dive in.
More granular user reporting
Our user reporting dashboard has now been split into two sections via tabs: labeler and reviewer. This helps you track the efficiency of your labelers and reviewers more closely and make more informed decisions around workforce planning and task assignments.
The labeler tab includes elements such as a role column to toggle information related to specific roles (owner, admin, manager, and labeler) and the ability to toggle by annotation type or object class. You can now also review project-level statistics such as labeling progress, average object count per label, and average time per label by labeler. In addition, a progress column has been added so that you can see the status of all labels, including submitted or skipped ones, at a glance. Finally, this table has been split between labeling and review results so you can quickly find the information most relevant to your current task.
The review tab, on the other hand, is similar in the sense that we’ve added the ability to toggle by role, as well as added additional such as project-level statistics (review progress and label requested for review per percentage) and a progress column for checking the status approved labels.
New SDK Functions
Since the last update, we’ve also worked to continue to build out the scope and scale of our SDK. This includes two new functions:
• Get exports
• Get data for image projects
If you’re looking for even more developer and shortcuts tools for the Superb AI platform, check out our recently released Recipes page. We’ll be adding more recipes to this space over the coming months, so be sure to check back often.
Additional Features and Improvements
• Hotkey shortcut (cmd+h) to hide/display all objects
• Hotkey shortcut (shift+m) to switch between edit and view modes
• Change the tracking ID for objects
• Map multiple classes of different annotation types to one common object AI class
• Object property count metrics in the analytics dashboard
• Improved performance for labels with many figures
Conclusion
That’s it for this month. However, we have many big, and we mean big, releases coming later this year that go beyond the labeling aspect of data preparation, so stay tuned for more to come! And as always, you can head over to our changelogs to get all the great nitty-gritty details about what’s changed. Or, if you haven’t given our platform a try yet, you can sign up for free to get started today.
