RectLabel
rectlabel.com
- Free trial
- macOS
What RectLabel is
RectLabel is an offline macOS application for annotating images used in object detection and segmentation. It supports polygons, pixels, bounding boxes, oriented boxes, cubic Bezier curves, line segments, points, and skeleton keypoints. Model-assisted labeling includes Segment Anything Model 3 text and box prompts, Segment Anything Model 2, Cellpose, and Core ML models such as RF-DETR and YOLO26. Automatic text recognition handles lines and words. Users can configure objects, attributes, and hotkeys, search image details in a gallery, convert video into image frames, and augment images. Exports include YOLO, COCO, CreateML, DOTA, CSV, mask images, and classification folders. The maker says image and annotation data are not collected, and RectLabel Pro has no network connections. RectLabel requires macOS 13.5 or later and is for Mac only. The RectLabel plan costs $9.99 a year after an untimed trial; RectLabel Pro costs $19.99, paid once.
Who it suits
RectLabel suits developers and dataset teams preparing detection or segmentation labels on Mac computers. It fits users who need varied annotation types, model assistance, and multiple export formats.
Good to know, and what to watch
- Supports many annotation shapes and keypoints
- Model-assisted labeling includes SAM and Core ML models
- Exports YOLO, COCO, CreateML, DOTA, and masks
- Offline operation and stated data privacy
- Untimed free trial available
- Mac only
- Requires macOS 13.5 or later
- No free plan
- RectLabel Pro requires a one-time $19.99 payment
Plans
Every plan and priceRectLabel
RectLabel Pro
RectLabel
The verdict
RectLabel offers broad annotation and export choices in an offline Mac workflow. Its operating-system requirement and paid license structure matter for teams choosing a labeling tool.
What the maker publishes
- Annotation types
- The app supports polygons, pixels, bounding boxes, cubic Bezier curves, line segments, points, oriented bounding boxes, and skeleton keypoints. source
- Compatibility
- The App Store listing says RectLabel requires macOS 13.5 or later and is for Mac only. source
- Data privacy
- The maker's privacy policy says it does not collect images or annotation data. source
- Export formats
- RectLabel exports to YOLO, COCO, CreateML, and DOTA, as well as indexed color and grayscale mask images. source
- Founded
- 2017 source
- Headquarters
- Tokyo, Japan source
- Integration
- For larger deployments, the maker recommends assigning app licenses to devices through Apple Business Manager managed licenses. source
- Maker history
- Ryo Kawamura says he is an independent full-stack engineer from Tokyo and released RectLabel's first version in 2017. source
- Models
- It supports polygon and pixel labeling with Segment Anything Model 2 and Cellpose, and automatic labeling with Core ML models including RF-DETR and YOLO26. source
- Network use
- The maker says RectLabel Pro has no network connections because it has no in-app purchases. source
- OCR
- The app provides automatic text recognition for lines and words. source
- Purpose
- RectLabel is an offline image annotation tool for object detection and segmentation. source
- SAM 3
- Text and box prompts from Segment Anything Model 3 can label multiple objects at once. source
- Search and workflow
- Users can configure objects, attributes, and hotkeys, search objects and image details in a gallery, convert video to image frames, and augment images. source
- Support
- The maker directs users to post problems to GitHub issues or email [email protected]. source
- Target users
- The maker describes itself as building simple, intuitive tools for developers and says RectLabel is for annotating images for object detection and segmentation. source
Checked against the maker's pages on 9 Oct 2026.