What is the best experience to use for a photo tagging feature? to view tagging when we hover on the photo? or having a separate button in the photo page that calls for tagging ?
2
-
When you say "tagging", do you mean tagging specific regions of an image (e.g. tagging faces on Facebook) or just associating keyword metadata with the image?– Matt ObeeCommented Oct 22, 2012 at 9:26
-
Yes tagging faces or items in a photo, we can say like the facebook tagging.– Jamila HyasatCommented Oct 22, 2012 at 9:33
Add a comment
|
2 Answers
The best photo tagging experience is no tagging: when the photos are just "magically" tagged correctly, either by a program or by someone else.
Then comes the first fallback. community tagging.
It depends on the context wether tagging mode should be enabled by default or not. I guess if showing already tagged items is on by default, then tagging mode should be default as well on desktop interfaces, where hover mode actually exists.
You could, for example, show a frame around untagged faces on hover and say:"Who's this?" instead of a person's name.
On more and more prevalent mobile and tablet (touch) interfaces however, there's no such thing as hover: here, accidental taps could happen, so tagging is better left off by default, except when the whole application is mainly about tagging.
But no matter which mode you choose, it's always recommended to make computers help people: even if a computer cannot recognize a face for sure, nowadays we can more or less recognize where is a face. Therefore the computers can pre-calculate the rectangles where it is likely that there's a face.
-
decent advice if you've got a killer server and a AAA-class programmer. Let's be real, face detection is not an easy algorithm even google and facebook fail to do it successfully at times. A small business would be hopelessly bad at the task. Magically tagged correctly is so laughably difficult that you might as well have said that the business can magically print money. Yes I am aware that google does have a face recognition program, but they also practically print money. Commented Oct 22, 2012 at 13:26
-
I apologize if my comment was rude, it's entirely possible that you might not have known the relative difficulty of the task you suggested. I come from a programming background, and people often suggest very difficult tasks (or sometimes impossible) as if they're trivial. Commented Oct 22, 2012 at 13:31
-
@VoronoiPotato We risk going off topic but, while it's difficult to do reliably in mission-critical situations, you can achieve very basic facial recognition (suitable for this use case) using client-side scripting. Although this is of course limited to "we think there's a face here" rather than "we think there's a person called Bob here". Commented Oct 22, 2012 at 13:52
-
1Recognizing that it is a face, yes, recognizing whose face, no. Not to mention that a poorly implemented facial recognition system can be a social faux pas. For example failing to account for varying skin tones, eye shapes, can be quite (unintentionally) offensive. The human face is pretty complex, and an intimate part of our social interactions. It's a tall order to expect a computer to handle the niche scenarios. It would not feel good if the machine couldn't recognize your face because of cleft palate. EDIT: I think we're [mostly] on the same page. Commented Oct 22, 2012 at 14:00
-
1@VoronoiPotato: I think we both agree, that the best experience is when these algos are there. After that, I started to enlist the "escape routes", mentioning recognizing simply that there is a face. I do agree it's a hard topic and it takes a lot of neural networks and perhaps it's nearly magic, but an OpenCV-based classifier is not unreacheable even for a startup (it detects just the existence of a face). Still I hold: the best experience is when it's not manual labor– AadaamCommented Oct 22, 2012 at 16:14
I think honestly it depends on the kind of website or program you're trying to run. If the tag depends on a specific subject in the image that the viewer could not otherwise identify such as a face, or a type of fruit that isn't common knowledge (etc) then I think the hover tagging is essential. Otherwise if it's to denote a general mood or general subject matter such as a Yosemite park or sorrowful, then the button is more helpful. You might find that your users value both, as the general tags can be very useful for sorting.
