Idanwo ti o sopọ mọ itọsọna · Alabọde Ipele

Idanwo Awọ Aworan

Checks why colorization is ambiguous, the Lab input/output channels, classification over color bins and learned image priors.

Jẹmọ itọsọna awọn ọnaAwọ Aworan
Ibeere 1 ti 8

Kilode ti awọ aworan ṣe ka iṣoro 'aisan'?

Awotẹlẹ gbogbo ibeere8 ibeere

Awọn idahun ati awọn alaye han bi o ṣe n mu ibeere naa.

  1. Kilode ti awọ aworan ṣe ka iṣoro 'aisan'?
    • Iwọn grẹy kan ṣoṣo le ṣe deede si ọpọlọpọ awọn awọ gidi ti o yatọ
    • Awọn aworan grẹy ko ni alaye nkan elo ninu rara
    • Awọn aworan awọ ti tobi ju lati ṣiṣẹ
    • Awọn kamẹra ko le ṣe igbasilẹ iwọn grẹy mọ
  2. Ni ọna aaye awọ Laabu olokiki, kini nẹtiwọọki n gba bi titẹ sii?
    • Aworan RGB ni kikun
    • Nikan L (lightness) ikanni
    • Nikan ikanni pupa
    • A ijinle map ti awọn ipele
  3. Kini idi ti Zhang et al. Asọtẹlẹ awọ fireemu bi ipin lori awọ 'bins' kuku ju ifasẹyin taara?
    • Nitori iyasọtọ jẹ ohun kan ṣoṣo ti awọn nẹtiwọọki nkankikan le ṣe
    • Lati jẹ ki awọn aworan kere
    • Lati yago fun aropin to wulo awọn awọ sinu ṣigọgọ, desaturated esi
    • Lati yọ iwulo fun data ikẹkọ kuro
  4. Bawo ni awọn awoṣe ṣe kọ ẹkọ kini awọn ohun awọ yẹ ki o jẹ laisi sọ fun wọn?
    • Wọn ti ṣe eto pẹlu ọwọ pẹlu atokọ ti gbogbo awọ ohun kan
    • Nwọn gboju le won laileto kọọkan akoko
    • Wọn ka metadata awọ ti a fipamọ sinu awọn faili greyscale
    • Wọn kọ ẹkọ awọn iṣaaju iṣiro lati awọn miliọnu awọn fọto awọ ti o yipada si iwọn grẹy
  5. Kini ipenija bọtini nigbati o ṣe awọ fidio dipo awọn fọto ẹyọkan?
    • Titọju awọn awọ ni ibamu lori awọn fireemu lati yago fun yiyi
    • Awọn fidio ko ni ẹya grẹy
    • Awọ ko ṣee ṣe lati ṣafikun si awọn aworan gbigbe
    • Kọọkan fireemu gbọdọ jẹ kan yatọ si awọ
  6. Kini orisun-apẹẹrẹ tabi awọ ti o da lori itọ jẹ ki olumulo kan ṣe?
    • Yipada awọn aworan awọ pada si iwọn grẹy nikan
    • Ṣe itọsọna iṣẹjade nipa lilo awọn aworan itọkasi tabi awọn itọka awọ ti a tẹ
    • Mu nẹtiwọki nkankikan ṣiṣẹ patapata
    • Tumọ aworan naa si ede miiran
  7. In the CIE Lab setup described in the guide, which channels does the network predict?
    • The L lightness channel only
    • RGB red and green but not blue
    • The a and b chrominance channels
    • The image’s camera pose
  8. Why can direct regression toward an average color produce dull colorization?
    • The model is forced to output only black
    • Regression removes the lightness channel
    • The network cannot use training examples
    • Several plausible colors for a gray pixel may average to a muted, less likely color