quiz bu lëkkale ak tegtal · Diggante Niveau

Quiz buy xàmmee jëf

Jàngaleel ni xeetu wideo yiy xàmmee jëf di jëfandikoo firndey jamono, misaali yëngu ak convolution spatiotemporal.

Yooni njiit yu jëm ciXamme Jëf
Laaj 1 de 8

Lan mooy wuutale xàmmee jëf ak xaaj nataal buñ miin?

Xoolaatal laaj bu nekk8 laaj

Sooy def quiz bi, tontu yi ak leeral yi dañuy feeñ.

  1. Lan mooy wuutale xàmmee jëf ak xaaj nataal buñ miin?
    • Dafay wax ci ni pixel yi di soppeekoo ci diir bi, du benn kaadar kese
    • Dafay jëfandikoo nataal yu gëna mag
    • Ci wideo yu weex ak yu ñuul kese lay dox
    • Soxlawolen ay done tàggat
  2. Ci reso buy xàmmee jëf bu am ñaari stream, lan la ñaareelu stream bi di faral di def?
    • Formu onde audio
    • Flux optik buy màndargaal yëngu
    • Mbind yu am solo
    • coordonnées GPS
  3. Lan la convolusioŋ 3D di yokk suñu ko méngale ak convolusioŋ 2D buñ miin?
    • chaine de couleur
    • Dayo nataal bu gëna rëy
    • Axe temporal (waxtu) suko defee filtre yi mëna jaar ci kaadar yu bari
    • Dayo lote bu gëna mag
  4. Lan mooy 'inflation' bi xeetu I3D bi di jëfandikoo?
    • Yokk leeralu wideo
    • Yokk yeneen diisaay ci anam wu mucc
    • Fichier wideo bi
    • Yaatalal diisaayu reso nataal 2D ci 3D di leen toppandoo ci diir bi
  5. Lan moo waral setu done Something-Something siiw ci xàmmee jëf?
    • Dafay forse model yi ñu xam yoon wi ci jamono, duñu wéeru ci seen contexte
    • Wideo baayima kese la am
    • Amul benn etiketu
    • Benn nataal lañu ko defaree
  6. Ban pexe mooy jàppale xeetu xàmmee jëf yi ñu jàngee ci wideo yu amul etiket?
    • Etiketu kadre par kadre
    • Modeling wideo maske buñuy saytu boppam
    • efaase bépp done buy yëngu
    • Jëfandikoo audio kese
  7. How many action-template classes did the initial Something-Something dataset paper describe?
    • 17 classes
    • 1,740 classes
    • 174 classes
    • 17,400 classes
  8. What pretraining strategy does VideoMAE use?
    • Training only on single labeled frames
    • Replacing video with a text transcript
    • Predicting optical flow without reconstructing masked content
    • Masking a high proportion of video tokens and reconstructing them from visible tokens