We carefully designed the UX of the annotation tool from the ground-up to make annotating as simple and efficient as possible.
Remo currently supports three types of annotations:
- Image level annotation for image classification tasks
- Bounding boxes for object detection tasks
- Polygons for segmentations tasks
Regardless of the annotation task selected, you can add tags to an image from the corresponding menu.
Tags are designed to give total flexibility in categorising images, and can later be used to index or search your images.
To facilitate managing the annotation process, you can assign an annotation status to your images. Annotation status can be either:
- Done indicated by a green mark
- On Hold indicated by a yellow mark
- To do not marked in any way
This is extremely useful to organise the annotation process, as you might want to go back to annotate some difficult pictures at a later stage (just use the sort by button).
Efficient Annotation tools¶
Following the approach defined in the paper Extreme click for efficient object detection, we implemented the efficient annotation method "xclick", which should yield a speedup of up to 5x compared to the traditional "draw a box" annotation.
To facilitate annotation, the annotation tool features a number of shortcuts to the most common functionalities.
|Key||What it does|
|Hold Space Bar||Grab tool|
|Mouse Wheel||Zoom in / out|
|B||Draw a box (once and return to previous tool)|
|D||xclick (once and return to previous tool)|
|P||Draw a polygon (once and return to previous tool)|
|Backspace||Delete selected object (no confirmation needed)|
|Numbers and other letters||Select a class to annotate or visualise annotations|
If you have any suggestions on the Annotation tool, accessibility issues or new features that you would like included, we are all ears. The best way to get in touch is to write on the forum.