Ijinle Ohunkohun Monocular Ijinle adanwo

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

1. Kini iṣiro ijinle 'monocular' tumọ si?
   • Ijinle asọtẹlẹ lati aworan kan
   • Lilo awọn kamẹra meji fun iran sitẹrio
   • Lilo sensọ LiDAR kan
   • Iwọn iwọn ijinle pẹlu awọn igbi ohun
2. Kini ilana akọkọ ti DepthAnything fun iyọrisi gbogbogbo ti o lagbara?
   • Ifi aami-ọwọ datasetiti didara didara kekere kan
   • Ikẹkọ lori aijọju 62 milionu awọn aworan ti ko ni aami aifọwọyi
   • Lilo nikan sintetiki game engine data
   • Gbẹkẹle awọn orisii kamẹra sitẹrio
3. Egungun ẹhin wo ni DepthAnything kọ lori?
   • A ResNet-18 CNN
   • Nẹtiwọọki LSTM loorekoore
   • Oluyipada iran DINov2
   • Oluwari YOLO kan
4. Iru ijinle wo ni DepthAnything atilẹba ṣe jade ni akọkọ?
   • Gangan ijinna ni awọn mita
   • Nikan alakomeji iwaju/lẹhin
   • GPS ipoidojuko
   • Ijinle ibatan (sunmọ si siwaju sii)
5. Bawo ni DepthAnything V2 ṣe ilọsiwaju awọn alaye ti o dara ati awọn egbegbe?
   • Nipa ikẹkọ olukọ lori data sintetiki pẹlu otitọ ilẹ pipe, lẹhinna distilling
   • Nipa fifi awọn aami alariwo gidi kun diẹ sii
   • Nipa yiyọ kooduopo
   • Nipa lilo awọn igbewọle iwọn kekere
6. Ohun elo wo ni kii ṣe lilo aṣoju ti awọn maapu ijinle monocular?
   • Aworan-ipo isale blur
   • Encrypting ijabọ nẹtiwọki
   • ControlNet ijinle karabosipo
   • 2D-si-3D Fọto iyipada
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