See all uploaded media waiting for review. Left sidebar lists files – click to load one into view. Item card contains:
-
Short file id
-
Upload user
-
Upload date
-
✅ Green = approved
-
❌ Red = rejected
-
🔵 Blue = untouched
-
🟣 Purple = duplicated
Use filters at the top to narrow by status, label, date or author.
Main canvas shows the media + label overlays.
Right sidebar holds the full label tree for that file along with file info such as:
- File id
- Validation user
- Validation date
- Label tree
- Manage buttons
You can:
- View the media, move it, zoom
- Edit labels in-place
- Add/remove object groups
- Accept (✔) / Reject (✖)
- Download item
Use these to streamline flow:
← / →— switch fileX— reset canvas stateA— accept currentD— reject current- Hover on thumbnail to preview, click ⬇️ to download.
Once all files are green or red you can go to the next page if some exists. This could be done only manualy by now.
This system improves data integrity, avoids duplication of annotation work, and improves the consistency of image usage throughout the dataset lifecycle.
The platform includes a built-in duplicate detection mechanism to streamline annotation workflows and prevents redundant effort.
Key Features
- Automatic Detection: Upon image upload or validation, the system checks whether the image already exists in the current project. Detected duplicates are flagged immediately.
- Visual Indicator: Duplicates are visually marked with a large
DUPLICATEtag displayed directly on the image view.
-
Best Quality Selection: When duplicates are found, the system automatically selects the best quality version (e.g., higher resolution) as the primary reference image. This version will be used upon downloading
-
Duplication Browser: A new "show duplicates" button has been added, allowing quick access to all known duplicates of a given image, side-by-side.
-
User Decisions: Annotators can manually resolve duplicates.
Duplicate checks currently operate within a single project only. and will be extended across all projects in the future


