Annotate & automate
Click to segment. Describe to detect.
Create and refine annotations with click segmentation, similar-object search, zero-shot models, and local or managed cloud labeling.
Use the GPU you already have
Click segmentation and similar-object search use WebGPU on compatible devices, with fallback support. Find similar objects within the same image and choose which results to keep.
Find objects with text
Use zero-shot detection to describe the objects you want to find. Zero-shot classification lets you provide your own candidate labels. Other local detection and classification models are also available.
Give the batch a head start
Run local models or managed cloud detection and classification over an image range. Set confidence thresholds, skip existing annotations, and track batch progress.
Keep the final decision
Edit boxes, points, and polygons in the canvas. Keep labels, saved annotations, and image navigation together as you refine the results.