Clarifai is proud to introduce version 6.6 of our platform, and the official public release of Labeler. Labeler makes it easy to take on labeling projects of any size, so that you can develop high quality training datasets for your AI-powered business solution. Plus we are introducing Collectors, so that you can easily pipe data into your app.
Labeler includes a ton of new features and functionality that help you label data accurately, affordably and quickly through AI augmentation.
We are supercharging our detection game with tools that make bounding box labels a snap. You can provide annotation within a single box-shaped region of an image or video. To use bounding box detection, you must start with a workflow that offers detection capabilities. From here you can label detected regions, or draw your own bounding boxes for labeling. Click here to learn more.
Polygon labels take detection-type models to the next level - enabling what is known as "semantic segmentation". With polygons labels you can outline the actual shape of the object that you would like to label, providing a whole new level of precision and accuracy. Try it yourself.
Interpolation allows you to quickly label multiple frames of video with the same concept. Interpolation will draw a series of bounding boxes that change size and location so that you can track the movement of an object in a video. Learn more.
Labeler provides special tools for working with images and video that have been designed by and for professional data annotators. You can enhance the visibility of your photos with image adjustments. Image adjustments can be combined. Powerful zoom and panning features enable you to closely inspect specific regions of an image. Just click reset to return to the original version of your input. Check it out.
Tasks enable you delegate labeling jobs to your team, and they let you scale up your labeling operations to take on projects of any size and complexity. Tasks include lots of helpful features that enable to you communicate with your team, divide and delegate work, and manage workforce performance. Try it out by creating your first labeling task.
Labeler provides smart tools for accelerating the data review and QA process that speed up the review and approval training datasets. AI automation and consensus reviews help you to efficiently spot check and QA large datasets and can speed up the process by an order of magnitude. You also have access to a helpful statistics dashboard that gives visibility to the performance of individual teams and team members. Learn more.
Collectors capture input data for your app. They enable you to pipe in data from production models automatically, and are the key to unlocking many platform training capabilities like active learning. Collectors are available with Essential and Enterprise plans to help you manage data ingestion at scale. You can create app-level collectors to monitor specific models and specify sampling rules for triggering data ingestion. Set up your own collector.