What is CVAT (Computer Vision Annotation Tool)?
Blog post from Roboflow
CVAT, or Computer Vision Annotation Tool, is a web-based, open-source image annotation platform initially developed by Intel and now maintained by OpenCV. It supports a range of annotation tasks, including object detection, classification, tracking, and segmentation, and offers various tools such as rectangles, polygons, and model-assisted labeling. CVAT allows users to manually annotate images or leverage AI models to automate the process, enhancing efficiency. The tool is accessible via its online interface or can be installed locally to overcome internet-based limitations such as task and data size restrictions. Users can export annotations in multiple formats suitable for training computer vision models like YOLOv4 and EfficientDet. CVAT is part of a broader ecosystem of annotation tools and is complemented by platforms like Roboflow, which simplifies data annotation through features like auto-suggested defaults and data augmentation.