Kediska Isbeddellada Deformable

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

1. Muxuu ku soo kordhinayaa is-daba-marin qallafsan marka loo eego is-bedbeddelka caadiga ah?
   • Cabbirka kernel go'an oo weyn
   • La baray 2D dhimis oo beddeleysa dhibic kasta oo muunad
   • Tallaabo isku urursi ka dib lakab kasta
   • Miisaanka aan tooska ahayn ee ka dhacaya
2. Sidee loo abuuraa muunad ka-soo-baxa is-beddelka?
   • Waxa gacanta ku sifeeyay injineerku
   • Waxay yihiin joogtoyo go'an oo laga soo bilaabo miiska raadinta
   • Waxa laga soo min guuriyay qaabkii hore
   • Lakabka convolution dheeraad ah ayaa iyaga ka saadaaliya gelinta
3. Sababtoo ah dhimista la saadaaliyay inta badan waa jajab, sidee baa qiyamka pixel looga soo qaatay boosaskaas?
   • Isdhexgalka laba-geesoodka ah
   • Dariska ugu dhow oo kaliya
   • Isku geynta ugu badan
   • Waa laga booday
4. Muxuu Deformable ConvNets v2 (2018) ku soo kordhiyay asalka?
   • Isku xirka soo noqnoqda
   • Khasaare cusub oo shaqeeya kaliya
   • Miisaanka qaabaynta hal-muunad ee cabbiraya wax ku biirinta dhibic kasta
   • Meesha laga saaray dhammaan wixii la dhimay
5. Waa maxay sababta is-beddellada qallafsan ay si gaar ah ugu caawiyaan walxaha aan caadiga ahayn?
   • Meesha ay ku qaabili karaan waxay u rogman kartaa si ay ula socdaan muuqaalada dhabta ah ee shayga
   • Waxay kordhiyaan xallinta sawirka
   • Waxay ku cadaadinayaan cabbirka qaabka
   • Waxay u beddelaan sawirada midabka midab cawl
6. Qaabkee ogaanshaha casriga ah ayaa khuseeya fikradda qallafsan ee dareenka beddelka?
   • ResNet-50
   • AlexNet
   • U-Net
   • Deformable DETR
7. For an N-point two-dimensional deformable kernel, how many offset components are predicted at each output location?
   • N: one class probability per point
   • N squared: one displacement for every pair of points
   • 2N: one horizontal and one vertical displacement per point
   • Two total, shared by all N points
8. Where does deformable RoI pooling apply adaptive offsets?
   • Only to the detector's output class names
   • To sampling locations within a proposed object's region features
   • To the ground-truth box coordinates before annotation
   • Only to the image file's compression blocks