Open source tool for collaborative image & video annotation, featuring predefined tags, bounding box interpolation, and review workflows
CVAT is an open source web-based tool for computer vision annotation and labeling of images, video, and other data. It allows users to draw bounding boxes, segment objects, track objects across frames, assign tags, and more. Some key features of CVAT include:
CVAT is implemented using modern web development stacks including Django, React, and Canvas. It can be self-hosted on an internal server or hosted in the cloud. The goal of CVAT is to provide a flexible and extensible computer vision annotation platform for teams to collaborate on dataset labeling and model training.