high priority request - upload screen- "anything"

[dobkeratops]

I would recommend adding a dummy label"anything" as the default choice ("upload an image that represents [anything]")

We have good weather now for taking photos.. time to be out and about, with a phone or even iPad. i can't easily use the site script at the minute have to find it again, transfer to desktop etc)... sorting out a good workflow for mobile users would be good. Just this tweak will make it easier gather photos.

next request would be the scene labels - about 50-100 ballpark IMO would be needed. these would then help you organise apush to 1000+ object labels , andgie good whole-image training signals eenbefore you annotate.. make the data as useful as possible as early as possible.

things like "garden vs park vs forest" all have similar objects, but in different proportions also street , town,village,city, residential street, hughstreet, highway, motorway, .. or docks svs harbour vs port vs marina.. riverside vs seaside, coastline etc. these words would be great to have

[bbernhard]

I would recommend adding a dummy label"anything" as the default choice ("upload an image that represents [anything]")

that should be easily doable. I guess I should be able to push that to production either tomorrow or the day after :)

sorting out a good workflow for mobile users would be good. Just this tweak will make it easier gather photos.

Yeah, definitely! If you have any ideas, please let me know :)

next request would be the scene labels

you are right, scene labels are mostly missing at the moment. But that's already the next big topic on my todo list. I hope that I can start working on that either this or the next weekend. The codebase should already support that, so I think it shouldn't be much work.

[bbernhard]

I've just pushed the changes to production. Not sure, if it was a good idea, but I've completely removed the labels dropdown from the donate page. The intention behind that was, that I would like to make the page as simple as possible - mainly because I think the donation page will be the "entry point" for most users to ImageMonkey. It should be dead simple for all users (even newcomers) to upload a snapshot without worrying about the right label.

For all the users that are more experienced I would like to provide two additional upload methods in the future:

• script based: basically the script that's already available, but more user friendly (better help text, commandline args,..etc)
• GUI based: images can be selected via image browser.

Both of those options allows the user to add labels. Additionally a user can specify that the uploaded images should be assigned to his personal image collection.

It's just an experiment, in case it turns out that removing the label dropdown was a bad idea, I can easily revert the changes again :)

[dobkeratops]

Not sure, if it was a good idea, but I've completely removed the labels dropdown from the donate page.

that will probably be ok.You can just make those images appear first in add labels mode, and concentrate on that UI for sorting out scenes, common co-occuring labels etc. The addition of one new feature makes another earlier feature kind of redundant. If you wanted them labelled immediately, you could perhaps even go directly into add labels mode after you drop a bush of images (or just put a shortcut there)

I saw the advanced label addition feature aswell, that's pretty interesting too (being able to access the materials etc). Deinitely lots of fronts on which to keep this tool interesting. If you see the brainstorming reference (use of AI in cgi) you'll see the kind of long term goals I have in mind..