Open bbernhard opened 6 years ago
basically if you wanted to keep the UI lean, IMO a universal browse/choose action mode (generalisation of existing 'explore' that includes option show info) would give you the most functionality; nonetheless, the current 'browse-and-annotate' is good, and by extention 'browse-and-label' could also be useful.
i guess this really depends on what is easier to actually do, but also think about discoverability for other users
At the moment I am thinking about introducing another mode (?mode=browse
) to the labels view - similar to the annotation view. I think in the end a unified browse based view is definitely something desirable. :) My hope is, that if we start with a completely separated browse based labels view first, that it helps us to gather some valueable information about how the unified mode could look like and what options we want to expose in the UI.
Ideally, we can keep the separated modes, even when we have the unified browse based mode in place. I think in some cases, the separated modes might even be superior to the unified mode. (the unified mode requires you to select what you want to do (validate, label, annotate), which means one additional click. In most cases that's perfectly fine, but if you want to bulk-annotate it's probably more convenient to use the separated mode)
now I recall add-labels has a goto ...
, so that would be quite close to a unified view already, just one extra click
@dobkeratops there is now a browse based label view at https://imagemonkey.io/label?mode=browse It allows you to search for label expressions (you can even include labels that are not yet productive in your search query). If you hover over an image, it shows you the labels.
edit: the autocomplete functionality is pretty slow and sometimes the browser hangs a bit (if I remember correctly, you already raised that issue a while ago). I'll look into that :)
great , i'll take a look .. I noticed the ability to label/annotate locked images too, thats great to have
... perfect! just tried road&~sky
.. to fill in missing sky labels (etc) - works very well. being able to click the common labels is handy too, i.e. this task can be done keyboard less.. scroll, click,click done,scroll..
the autocomplete functionality is pretty slow and sometimes the browser hangs a bit
right, it might be ok to disable it if it isn't possible to speed up; another suggestion (if the volume of completions matters?) would be to restrict autocomplete to the known label set. Common labels can be cut-paste; I think the default user view uses a drop-list of the known set anyway. (as I mentioned the common labels area does save me from using it)
proposed by @dobkeratops in #157