Ikibazo gihuza ibibazo · Hagati Urwego

Ubujyakuzimu Ikintu cyose cyimbitse ya Monocular

Ongera Ubujyakuzimu Ikintu cyose cyumwarimu-umunyeshuri moteri yamakuru, kodegisi ya DINOv2, ibisubizo byimbitse byimbitse, hamwe niterambere-ryamakuru muri V2.

Inzira zijyanye no kuyoboraUbujyakuzimu bwa Monocular
Ikibazo 1 Bya 8

Kugereranya ubujyakuzimu 'monocular' bisobanura iki?

Reba buri kibazo8 ibibazo

Ibisubizo nibisobanuro bigaragara mugihe ufata ikibazo.

  1. Kugereranya ubujyakuzimu 'monocular' bisobanura iki?
    • Guteganya ubujyakuzimu uhereye ku ishusho imwe
    • Gukoresha kamera ebyiri mubyerekezo bya stereo
    • Gukoresha sensor ya LiDAR
    • Gupima ubujyakuzimu hamwe nijwi ryamajwi
  2. Ni ubuhe buryo bukomeye Ikintu cyose cyingenzi cyo kugera kuri rusange?
    • Kurambika intoki ntoya ntoya yo mu rwego rwo hejuru
    • Amahugurwa agera kuri miliyoni 62 yimodoka yanditseho amashusho atanditse
    • Gukoresha gusa moteri yimikino yimikino
    • Kwishingikiriza kuri kamera ya stereo
  3. Ni ubuhe buryo bw'umugongo kodegisi Yimbitse Ibintu byose byubaka?
    • ResNet-18 CNN
    • Umuyoboro wa LSTM usubiramo
    • Guhindura icyerekezo cya DINOv2
    • Ikimenyetso cya YOLO
  4. Ni ubuhe burebure bwimbitse Umwimerere Ikintu cyose gisohoka cyane?
    • Intera nyayo muri metero
    • Gusa binary imbere / inyuma
    • GPS ikora
    • Ubujyakuzimu bugereranije (hafi na kure)
  5. Nigute Ubujyakuzimu Ikintu cyose V2 cyateje imbere amakuru arambuye?
    • Muguhugura mwarimu amakuru yubukorikori hamwe nukuri kwuzuye, hanyuma akabeshya
    • Mugushyiramo ibirango byuzuye urusaku
    • Mugukuraho kodegisi ya transformateur
    • Ukoresheje ibisubizo-bito byinjira
  6. Ni ubuhe buryo butari uburyo busanzwe bwo gukoresha ikarita yimbitse?
    • Igishushanyo-cyerekana imiterere yinyuma
    • Guhishira urujya n'uruza
    • IgenzuraNet yimbitse
    • Guhindura amafoto 2D-kuri-3D
  7. What dense prediction head does the Depth Anything guide pair with its DINOv2 encoder?
    • A Marching Tetrahedra mesh extractor
    • A CLIP text-image similarity head
    • A DPT-style dense prediction head
    • A causal language-model decoder
  8. How does Depth Anything V2 use a teacher model with unlabeled real photos?
    • The teacher converts each photo into a 3D mesh
    • The teacher supplies only image captions to a language model
    • The teacher replaces all synthetic depth labels with manually measured distances
    • The teacher generates pseudo-depth labels that are used to train student models