CVAT

CVAT (Computer Vision Annotation Tool) is an open-source, web-based tool for annotating images and video for computer vision model training — drawing bounding boxes, polygons, keypoints, and segmentation masks, with support for semi-automatic annotation using pretrained models.

It's built for teams annotating data at real scale — multi-user projects, task assignment, quality review workflows, and format export compatible with major ML frameworks — rather than a single-person quick-labeling tool.

Originally developed by Intel, now maintained independently; a common choice for teams building custom object-detection or segmentation datasets who want a self-hosted, no-per-image-cost alternative to commercial annotation platforms.