[{"data":1,"prerenderedAt":12},["ShallowReactive",2],{"stack-tr-en-cvat":3},{"tagline":4,"description":5,"strengths":6,"useCase":11},"The open-source, self-hostable annotation tool for building image and video datasets.","CVAT is a web-based annotation application, released under the MIT license, that you deploy on your own servers with Docker Compose or use as a hosted service. It covers boxes, polygons, points, cuboids, masks and skeletons, interpolates between keyframes to annotate video, and speeds up the work with semi-automatic annotation backed by models such as SAM or YOLO. Projects, tasks, jobs, review and quality control organize a team of annotators, and exports go out as COCO, YOLO, Pascal VOC or Datumaro. It's our choice when images must not leave the client's infrastructure.",[7,8,9,10],"Boxes, polygons, points, cuboids, masks and skeletons on images and video.","Video: keyframe interpolation and object tracking to annotate fewer frames.","Semi-automatic annotation: SAM, YOLO or your own model plugged in as a serverless function.","Self-hosted, MIT license, REST API and Python SDK, COCO, YOLO and Pascal VOC export.","Building detection or segmentation datasets, annotating surveillance video, confidential industrial data.",1792675980051]