sadly, this feature is not yet built in.
you could try to hack the tool by say: read in the annotation file, skip though already seen images, continue where you stopped, and append to annotations.
but for now, i'd recommend a much simpler approach:
- look at your current annotations file, find out, which was the last parsed image, then:
- move all already seen images into a temp folder (so you won't annotate them again)
- start tool again with a new annotation file, and work on remaining images.
- once you're finished, concatenate the 2 annotation files, and move images back to original location