ImageMonkey / imagemonkey-core

ImageMonkey is an attempt to create a free, public open source image dataset.
https://imagemonkey.io
47 stars 10 forks source link

donate: maybe allow unlabled donation -by default #217

Open dobkeratops opened 5 years ago

dobkeratops commented 5 years ago

my thinking is: given the Add Labels screen - which is very relaxing and easy to use - it would be better to make the act of uploading an image take no thought.. just click the tab and drop the image, without even having to think about a labels. The wording could be something like "upload an image: default label: ". You could still retain the ability to set a default label.

eg some people might happen to have interesting photos , or be in interesting places, and they dont want to spend time labelling on the site... whilst others are stuck inside in winter or whatever with time on their hands :) the beauty of online collaboration is the combination. Or it could even be the same person , and they're just squeezing the activity into their day at different times.

Second to this you could prioritise serving unlabelled images in the "add labels" mode.

This might also work better for any forthcoming unified mode,

.. and it would also allow submitting images for label suggestions .. actively suggest ways to broaden the vocabulary..

bbernhard commented 5 years ago

You are right, that definitely makes sense! I think it should be easy to change that.

Second to this you could prioritise serving unlabelled images in the "add labels" mode.

I have to check again, but I think it should already be the case. But it's possible that the recent extensions broke the functionality..I'll have a look :)

btw: I am also working on a small GUI application which should it make easier to extract frames from a video. It's inspired by your experiments with VLC(not sure if I remember that one correctly, but I think you extracted the frames with VLC, right?).

Here's a short demo:

frames_extractor

It's still work in progress, but it hopefully gives you a rough idea of what it will look like. The basic workflow will be:

dobkeratops commented 5 years ago

btw: I am also working on a small GUI application which should it make easier to extract frames from a video. It's inspired by your experiments with VLC(not sure if I remember that one correctly, but I think you extracted the frames with VLC, right?).

Interesting. All I did was grabbed them from a videoplayer, it was ok with trackpad scrolling to advance/rewind and the screen grab hotkey - but thumbnails in a GUI tool as you show would be better. Anything to help you pick out varied images (regular sampling of traffic footage would have a lot of very same-y frames)