To catch click events and to have a smooth experience with the ui.image annotator, we would like to have something like a point selector that reacts on click events and acts like a trigger. This way we can prompt models given an image point coordinate to obtain the polygon vertices which we would like to use to draw the new polygon shape with the ui.image_annotator.
Describe alternatives you've considered
It's also possible to prompt models of interest with a box but this would slow down the data labeling process and would not be as smart and efficient as using the point coordinate.
Requested solution
To catch click events and to have a smooth experience with the ui.image annotator, we would like to have something like a point selector that reacts on click events and acts like a trigger. This way we can prompt models given an image point coordinate to obtain the polygon vertices which we would like to use to draw the new polygon shape with the ui.image_annotator.
Describe alternatives you've considered
It's also possible to prompt models of interest with a box but this would slow down the data labeling process and would not be as smart and efficient as using the point coordinate.