Auto Annotation streamlines the labeling process on Foundry by leveraging state-of-the-art models,
significantly reducing manual effort and potential for human error.
We identified several challenges in manual annotation:
Laborious and monotonous task execution
Higher potential for human error
Increased number of Issues during review stage —> Time-consuming correction processes
Reduced annotation efficiency at scale
These challenges become particularly significant when dealing with thousands of frames,
making automated assistance essential for maintaining both speed and quality.
Autolabel specific frames right on the Canvas using the new Agent tool
Both approaches leverage classical machine learning methods and state-of-the-art models (Grounding Dino, Grounding SAM) and
provide an intuitive interface for annotation review and adjustment.