AnnoLoom
A modern desktop image annotation tool for computer vision datasets
Bounding boxes, polygons, and keypoints — in one clean PyQt6 app.
Install
pip install annoloom
Run
annoloom
Then Open Folder and point it at a directory of images. Draw boxes, polygons, or keypoint sets, assign labels, and save.
AnnoLoom UI
AnnoLoom Annotation Marking
Features
- Three shape types: bounding boxes, polygons, and labeled keypoints (with optional skeleton connections for pose-style annotation).
- Undo / Redo: Ctrl+Z to undo, Ctrl+Shift+Z to redo — up to 50 steps.
- Zoom in / Zoom out: Ctrl+Scroll wheel, or toolbar buttons. Zoom to Fit with Ctrl+0.
- Keyboard shortcuts:
A/D— previous / next imageB— switch to Box toolP— switch to Polygon toolK— switch to Keypoints toolEnter— finish polygon or keypointsDelete— delete the selected shapeEscape— cancel in-progress shapeCtrl+S— saveCtrl+Z/Ctrl+Shift+Z— undo / redo
- Auto-save: annotations are saved automatically when you switch to the next image — no more lost work from forgetting Ctrl+S.
- Annotation count: the shape panel header shows how many shapes are on the current image, e.g. "Shapes on this image (3)".
- Dark / Light theme toggle: switch between dark and light UI from the toolbar.
- Dataset Statistics: one-click summary — total images, annotated vs unannotated count, total annotations, and per-class breakdown.
- DrawAnnote / DelAnnote workflow: click DrawAnnote, draw a shape, it auto-finishes; select a shape in the list and click DelAnnote (or press Delete) to remove it.
- White crosshair guide lines follow your cursor while drawing, to help line up box edges precisely.
- Auto-fit, centered canvas: images scale to fill the available space and stay centered.
- Natural, numeric image ordering: files like
img2.jpg,img10.jpg,img100.jpgare listed in true numeric order. [current / total]counter in the title bar, e.g.[3 / 500].- Custom YOLO class numbers: pin any number to a label for YOLO export.
- Big, unmissable "✓ Labels saved successfully" banner on every save.
- Native save format (
<image>.aloom.json): lossless. - Export formats:
- YOLO (
.txt, normalized box coordinates, custom class numbers) - Pascal VOC (
.xml) - COCO (
.json, supports boxes, polygon segmentations, and keypoints)
- YOLO (
- No default "object" class: the label field starts empty so you type only the classes you actually need — no stray "object" category appearing in your exports.
Keyboard Shortcut Reference
| Key | Action |
|---|---|
A |
Previous image |
D |
Next image |
B |
Box tool |
P |
Polygon tool |
K |
Keypoints tool |
Enter |
Finish polygon / keypoints |
Delete |
Delete selected shape |
Escape |
Cancel drawing |
Ctrl+S |
Save |
Ctrl+Z |
Undo |
Ctrl+Shift+Z |
Redo |
Ctrl+= |
Zoom in |
Ctrl+- |
Zoom out |
Ctrl+0 |
Zoom to fit |
Ctrl+Scroll |
Zoom in / out |
Data model
Internally every shape is stored in a small dataclass model
(annoloom.core.shapes) at full float precision in image-pixel space.
Development
git clone https://github.com/Hakai99/ANNOLOOM
cd annoloom
pip install -e .
python -m annoloom.app
Building a standalone .exe (Windows)
build_exe.bat
This installs PyInstaller if needed and builds using AnnoLoom.spec.
Your executable will be at dist\AnnoLoom.exe.
License (OpenHands)
This project is licensed under a custom license — free to use, but modification and redistribution of modified versions are not permitted. See the LICENSE file for full terms.
Metadata
Release files for annoloom 0.1.3
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| annoloom-0.1.3.tar.gz | 664.2 kB | Without provenance | Details for annoloom-0.1.3.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| annoloom-0.1.3-py3-none-any.whl | Python 3 | none | any | Details for annoloom-0.1.3-py3-none-any.whl |
Total release size: 1.3 MB
Release files / annoloom-0.1.3.tar.gz
| Download URL | annoloom-0.1.3.tar.gz |
|---|---|
| Size | 664.2 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
898b55a7b38dcc23cb7c153694b4b0198660116afb54ff883dd42eb56f71536c
|
|
BLAKE2b-256 checksum How to use checksums |
0eda984867a82dc6846cf34424b0b69c82b959c2bac4a7d2fbef6afd77c068b3
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.10.11
|
Release files / annoloom-0.1.3-py3-none-any.whl
| Download URL | annoloom-0.1.3-py3-none-any.whl |
|---|---|
| Size | 662.3 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
660aa9d708b877d30f4e8bae5c3a4dceda8ab24e03f9a652937d68564b718e53
|
|
BLAKE2b-256 checksum How to use checksums |
15a3c187f31212337facb7db567568d54530ce603f51291319672ace8dd88ee6
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.10.11
|
Security for annoloom 0.1.3
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.