Dataloop is a data annotation platform for all kinds of labeling. An updated version of the software was released on June 6. This version featured improvements in its annotation studio and a new automation service.
Data annotation is a process of labelling data in different formats: pictures, videos and audio. This labelling helps machine learning algorithms make sense of raw data. It has helped to improve AI-based technologies like Computer Vision (for self driving cars) and AR/VR.
The annotation list in Studio 1.0 has been improved for a better user experience, while all functionality remains the same. Applications such as polygon-to-mask are now grouped in a single-action button.
Videos can now be annotated using Cuboid annotations – a kind of semantic segmentation. Also, the software’s Semantic Segmentation controls now have a new opacity slider and annotators are now able to toggle Pixel Smoothing. Annotators can also pan the screen using the middle mouse button.
On the automation side, dataloop released a new FaaS application center FaaS (function-as-a-service) 3.0. Easily install services, add functions and rerun executions to enhance your Dataloop experience.