Laajte ci Flow Optik

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

1. Lan mooy flow optik bi?
   • Pixel yiy yëngu ci diggante kaadar yu toppalante
   • Tamperatii melo bi ci kaadar bu nekk
   • Limu mbir yi ci benn nataal
   • Formu onde audio bi ànd ak wideo
2. Lan mooy 'constancy lumière' bi ñuy jëfandikoo ci flow optik bu yàgg bi?
   • Kadre bu nekk dafa wara am leeraay bu yam
   • Benn poñ buy toxu dafay tëye lu tollu ci benn leeraay ci biir kaadar yi
   • Leer gi dafa wara yokk bu baax ci kaw wideo bi
   • Wideo yu weex ak yu ñuul kese lañu mëna jàngat
3. Which method uses a 4D correlation volume and recurrent updates for optical flow?
   • Teserakt
   • Mask R-CNN
   • raft
   • Swin patch merging
4. Why can classic optical-flow assumptions fail in fast motion?
   • Because frames contain no pixels
   • Because flow is a text-only task
   • Because all motion is uniform
   • Large displacement, occlusion, or brightness changes violate local smoothness or brightness-constancy assumptions
5. Ban aplikaasioŋ moo sukkandiko ci xayma yëngu bi ci diggante kaadar yi?
   • Interpolaasioŋ kadre wideo ngir mouvement bu yeex
   • Saytu ortograafi ab këyitu bind
   • Tànnal nataal yi ci dayo fichier bi
   • Soppi kàddu ci mbind
6. Luy jafe-jafe 'oklusioŋ' ci debit optik?
   • Su wideo bi leer lool
   • Su pixel bi nëbbee wala feeñ, kon amul benn méngoo bu baax diggante kadre yi
   • Sudee viteesu kadre bi 30 fps
   • Suñu boole ñaari wideo
7. What is the output of dense optical flow?
   • One class label for the whole clip
   • A 3D mesh with metric scale
   • A displacement vector estimate for each image pixel
   • A transcript of the audio
8. What is one reason flow can be uncertain in a textureless region?
   • Every pixel is assigned a unique object name
   • The region has too many labels
   • Optical flow only works on text
   • There may be too little local visual information to identify a unique match