2 ms·

I am very surprised it got the color of the motorcycle wrong, it seems like the easiest thing to detect...

by Trufa 12y ago

Xophmeister 12y ago
It may not have a sophisticated enough vocabulary to distinguish 'pink' when 'red' was close enough. This effect is manifest in human languages which classify colours differently: say, for example, a language may have no word for 'blue', so the sky is 'green' to its speakers; it's still perceptually different to them, of course, but the lack of fidelity means it can't be communicated better than "sky green" or "grass green".
- Trufa 12y ago
  I understand there are a lot of subtleties, but it's still surprising for me that it can recognize a parked motorcycle from an awkward angle but can't distinguish normal english pink from red. Just to be clear, I don't mean it as a criticism, it just seems to be the easier part.
  - ajuc 12y ago
    There was no pink motorcycle in the training dataset most probably, and neural net may have failed to generalize colors beetween the objects. Similar error - yellow passanger car is described as yellow school bus - school bus is more common in yellow color.
- DanBC 12y ago
  Above the image of the pink (but labled red) motorbike there is an image of a child wearig a wooly hat. The hat is red with white fluff. That hat is labled pink. So, it knows "pink". And the motorbike isn't borderline pink / red -- it's not like hunting pinks -- it is definitely pink. Having said all that I'm amazed at the results. It feels like I'm living in the future.
jessriedel 12y ago
Much more likely than Xophmeister's explanation, I think, is that the brightness of a color in an image is relative to the lighting condition. (Remember that pink is just white mixed with red.) See image B: http://www.huevaluechroma.com/pics/3-4.jpg http://www.huevaluechroma.com/pics/3-4.jpg This is also true to some extent with the actual hue (i.e. red versus green, rather than brightness; see image C) but less so.

I am very surprised it got the color of the motorcycle wrong, it seems like the easiest thing to detect...

by Trufa 12y ago

I am very surprised it got the color of the motorcycle wrong, it seems like the easiest thing to detect...

It may not have a sophisticated enough vocabulary to distinguish 'pink' when 'red' was close enough. This effect is manifest in human languages which classify colours differently: say, for example, a language may have no word for 'blue', so the sky is 'green' to its speakers; it's still perceptually different to them, of course, but the lack of fidelity means it can't be communicated better than "sky green" or "grass green".

Trufa 12y ago
I understand there are a lot of subtleties, but it's still surprising for me that it can recognize a parked motorcycle from an awkward angle but can't distinguish normal english pink from red. Just to be clear, I don't mean it as a criticism, it just seems to be the easier part.
- ajuc 12y ago
  There was no pink motorcycle in the training dataset most probably, and neural net may have failed to generalize colors beetween the objects. Similar error - yellow passanger car is described as yellow school bus - school bus is more common in yellow color.
DanBC 12y ago
Above the image of the pink (but labled red) motorbike there is an image of a child wearig a wooly hat. The hat is red with white fluff. That hat is labled pink. So, it knows "pink". And the motorbike isn't borderline pink / red -- it's not like hunting pinks -- it is definitely pink. Having said all that I'm amazed at the results. It feels like I'm living in the future.

jessriedel 12y ago

Much more likely than Xophmeister's explanation, I think, is that the brightness of a color in an image is relative to the lighting condition. (Remember that pink is just white mixed with red.) See image B: http://www.huevaluechroma.com/pics/3-4.jpg http://www.huevaluechroma.com/pics/3-4.jpg This is also true to some extent with the actual hue (i.e. red versus green, rather than brightness; see image C) but less so.