Download LabelImg – Image Annotation Tool for Machine Learning Datasets

LabelImg is a free, open-source graphical image annotation tool designed for labeling object detection datasets used in training machine learning and computer vision models. It provides a simple, intuitive interface for drawing bounding boxes around objects in images and saving the annotations in popular formats like YOLO and Pascal VOC XML.

The tool is widely used in data science, AI research, and computer vision projects to prepare training datasets for models like YOLO, Faster R-CNN, SSD, and other object detection architectures. With LabelImg, annotators can quickly open a folder of images, draw labeled bounding boxes on each one, and export the annotations in a format ready for training frameworks like PyTorch, TensorFlow, and Darknet.

LabelImg supports keyboard shortcuts for all major actions, enabling fast annotation workflows when working with large datasets. It is cross-platform (Windows, macOS, Linux), written in Python with PyQt5, and can be run as a standalone executable or installed via pip.

Key Features

• Bounding box annotation – Draw rectangular bounding boxes around objects and assign class labels quickly.
• YOLO and Pascal VOC format – Save annotations directly in YOLO (.txt) or Pascal VOC XML format for major training frameworks.
• CreateML JSON export – Also supports Apple CreateML JSON annotation format.
• Folder batch workflow – Open an entire image folder and navigate through images one by one with auto-save.
• Keyboard shortcut support – Hotkeys for draw, save, next/prev image, and label switching to maximize annotation speed.
• Class label list – Define and save a custom class list for your project; labels persist across sessions.
• Cross-platform – Runs on Windows, macOS, and Linux; installable via pip or as a standalone binary.

How to Install

1. Install via pip: pip install labelImg then launch with labelImg from the command line.
2. Alternatively, download the standalone Windows executable from the link below — no Python installation required.
3. Launch LabelImg and open a directory of images via File > Open Dir.
4. Select your annotation format (YOLO or PascalVOC) from the toolbar, draw bounding boxes with the “Create RectBox” tool, and assign class labels.
5. Press Ctrl+S (or enable auto-save) after each image and navigate to the next with D key — annotation files are saved alongside your images.