AnyLabeling is an AI-assisted image annotation tool that combines the simplicity of classic labelers like LabelImg and Labelme with automatic labeling powered by YOLO and the Segment Anything family. Instead of drawing every box and polygon by hand, you let a model propose annotations and simply correct them.
AnyLabeling is open source on GitHub with about 3,500 stars and active development in 2026. It supports polygon, rectangle, circle, line and point annotation, plus text detection, recognition and key-information-extraction labeling for documents. Auto-labeling models include YOLOv8 for detection and SAM, MobileSAM, SAM 2/2.1 and SAM 3 for promptable segmentation, including open-vocabulary segmentation from text prompts. Hardware acceleration is available through CUDA, CoreML, DirectML, OpenVINO and vendor NPU providers via ONNX, so interactive labeling stays fast even on large images. The interface is available in English, Vietnamese and Chinese, and downloads and documentation are provided on the project site. Annotations can be exported in the common formats used by training pipelines, and batch auto-labeling over a whole folder makes it practical to bootstrap a first dataset in hours rather than days of manual drawing. Keyboard shortcuts and a modern interface keep high-volume annotation sessions manageable for professional labeling teams.
AnyLabeling is free and open source under the GPL-3.0 license. Download it from the project site or build it from source at no cost.
ML engineers, data teams and researchers who label image or document datasets and want model-assisted annotation to cut manual labeling time dramatically.