quiz bu lëkkale ak tegtal · Diggante Niveau
Man-man Reseau Piramid Quiz
Natt FPN ci kaw-wacc ak fusion man-man lateral, niveau-specific, upsampling dëkkandoo bi gëna jege, ak jëfandikoo ci gis-gis mbir yu bari.
Laaj 1 de 8
Ban jafe-jafe bu mag la Reseau Pyramid Feature di njëkka saafara?
Xoolaatal laaj bu nekk8 laaj
Sooy def quiz bi, tontu yi ak leeral yi dañuy feeñ.
- Ban jafe-jafe bu mag la Reseau Pyramid Feature di njëkka saafara?
- Gis mbir yu feeñ ci eskaal yu wuute lool
- Wàññi limu nataali tàggat yiñ soxla
- Di dindi ñaari boyet yuñ tënk ginaaw biñu leen gisee
- Modèle yuy kompresse ngir jëfandikoo mobile
- Buñu ko méngale ak tabax piramidu nataal (soppi dayo bi ak doxalaat reso bi), FPN moo gëna am njariñ ndax:
- Soxlawul benn reso backbone
- Jëfandikuwaat man-man yiñ jota xayma ci biir reso bi, di sakkanal xayma
- Dafay dindi bépp njariñu convolusioŋ
- Dafay dox ci nataal yu dóomu taal kese
- Ci FPN, lan mooy wareefu yoonu kaw wàcci?
- Dafay sànni màndarga yu am resolusioŋ bu woyof
- Dafay soppi nataal yi ci vecteur yu am solo ngir xaaj kese
- Dafay yor ay màndarga yu am solo, yu xóot ba ci dayo yu gëna kawe
- Dafay def suppression bu amul maximum
- naka lañuy boolee lëkkaloo yi ci wet yi ak man-mani upsampled yi ci kaw wàcci?
- Concatené chaine-wise ginaaw ga xaaj
- Yokkateg elemen ginaaw ab softmax
- Dañu ko dindi ngir xayma kàrtu wuute
- Yokk nañu ko ci wàllu elemen ginaaw bi ñu defee benn convolution 1x1 ngir méngale chaine yi
- Lan moo waral ñuy jëfandikoo ab convolution 3x3 ci bépp kàrtu màndarga buñ boole ci FPN?
- Ngir nooyloo kàrt bi ak wàññi pseudo ci upsampling
- Wàññi dayo spatial bi ci genn-wàll
- Ngir xaaj nataal bi ci kategori
- Ngir soppi man-man yi ci coordonnées boîte de limitation ci saasi
- Ban architecture moo yokk FPN ak fusion bidirectionnel buñ mëna jàng, pondéré?
- AlexNet
- BiFPN, jëfandikoo ci Det
- Lenet-5
- Word2Vec
- How is a top-down FPN feature map typically resized before lateral fusion?
- Downsampled to a single pixel
- Rotated 90 degrees
- Upsampled by a factor of two using nearest-neighbor interpolation
- Converted into a text sequence
- What does an FPN preserve by merging shallow high-resolution features with deeper semantic features?
- Only the original image colors
- Only the deepest layer’s low-resolution grid
- The exact training labels without feature extraction
- Fine spatial detail together with stronger semantic information