anylabeling

Why we included this project

Anyone assembling a computer-vision training set will save real time with AnyLabeling, a desktop annotator that pairs a conventional polygon, rectangle, circle, line and point editor with model-assisted auto-labeling. Instead of tracing every mask by hand, you can let YOLOv8 pre-label detections, then refine or generate instance masks through the Segment Anything line of models (SAM, SAM 2, SAM 2.1, SAM 3 and MobileSAM) with a couple of clicks. It also covers text detection, OCR and key-information extraction, which helps when your pipeline mixes vision with document understanding. Models download automatically on first use from Hugging Face, and the app runs either as a standalone build or a pip install, so small teams can get going without much setup. Teams coming from LabelImg or Labelme will recognize the workflow, since the interface is built on that familiar ground.