LabelLens

Crop, label and build datasets — sign in to continue.

LabelLens
0 / 00%

Images 0

Pick a folder to start

Choose a source folder with images and an output folder for your labelled data.

– / – Annotated

Adjustments are baked into everything you save. Double-click a slider to reset it.

On this image 0

Quick labels keys 1–9

Used in this output

Nothing saved yet.

Every save is logged. Revert removes the saved file(s) and sends you back to that image.

File format

Crop shape

Resize saved crops

Workflow

Choose folder

Drop images here to upload into this folder

Export dataset

Label foldersCrops & classified images, one folder per label (ImageNet / Keras style) + manifest.csv YOLOBounding boxes · images/ + labels/*.txt + data.yaml (Ultralytics) YOLO segmentationPolygons · normalized point lists per object COCO JSONBoxes + polygons in one annotations.json (Detectron2, MMDetection) Pascal VOCOne XML per image with bounding boxes CSVOne row per object — easy to open in Excel / pandas

How it works

Crop

Drag a box, type a name, press Enter. The crop is saved to output/<name>/ and the next image opens. Shift+Enter saves and stays, for several crops per image.

Classify

Type a label (or press 1–9) and Enter: the whole image is copied to output/<label>/.

Boxes

Draw boxes around objects and name each one. Enter saves the image with its boxes for YOLO / COCO / VOC export.

Polygons

Click points around an object's outline; click the first point or press Enter to close it. Used for segmentation.

←→ / ADPrevious / next image
EnterName selection / save
Shift+EnterSave crop, stay on image
EscCancel selection
Ctrl+ZUndo shape / revert last save
1–9Use quick label
DelDelete selected shape / last point
TabAutocomplete label
Space+dragPan (or right / middle drag)
WheelZoom at cursor
F / 0Fit to screen
R / Shift+RRotate right / left
GToggle black & white
MSwitch mode
NJump to next un-annotated
?This help