Dataset management: find, sort and split your visual data.
Images and video in one place. Folders, metadata and tags, filters you can combine, bulk actions, splits and versions.
Image dataset management, before you label and train.

Description, source and status on every folder, with Analytics and Build a Model.

A description and tags on any image, video or folder. Filter by tag anywhere.

Status, label, labeler and tags, or advanced filters with AND / OR.

Select everything a filter matches. Move, copy, assign tasks or delete.

Split by percentage, clone a folder, or take a random sample.

Save the draft as V1, V2, V3. Every model points to its version.
Data management questions.
Which files can I upload?
Images (PNG, JPG, BMP, TIFF) and videos (MP4, MOV, AVI, WMV, FLV, WEBM), plus ZIP files and COCO, Pascal VOC or YOLO label files.
What metadata can I add?
A description and tags on any image, video or folder. Tags become a filter for the whole repository.
Can I make a train and hold-out set?
Yes. Split a folder by percentage into two new repositories, clone it, or take a random sample. Splits need at least 10 items.
What happens to old versions when I delete?
You choose: delete from the version you're on only, or from every version.
Bring your visual data. Start free.
Free plan, no credit card. Bring your own images or start from a sample.