Kedis ku xidhan hagaha · Dhexdhexaad Heerka

Kediska Fogaanta Bilowga Fréchet

Dib u eeg FID's sifa-goob isbarbardhigga Gaussian iyo muunad-cabbir xasaasiyadeed, oo lagu daray beddelka qiimaynta-sawir-saacadnimada iyo kala duwanaanta.

Wadooyinka haga ee la xidhiidhaFogaanta Ka-hortagga Fr Chet
Su'aal 1 ee 8

Maxay muujinaysaa dhibcaha FID hoose?

Horudhac su'aal kasta8 su'aalo

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

  1. Maxay muujinaysaa dhibcaha FID hoose?
    • Tirakoobka sawirada la sameeyay ayaa ku dhow kuwa sawirada dhabta ah
    • Sawirada la sameeyay ayaa ah kuwo aan macquul ahayn
    • Qaabka ayaa loo tababaray waqtiyo ka yar
    • Sawiradu waxay leeyihiin xallin sare
  2. Shabakadee hore loo tababaray ayaa FID-da caadiga ah u isticmaashaa soo saarista sifooyinka muuqaalka?
    • ResNet-50
    • Bilawga-v3
    • VGG-16
    • BERT
  3. Sidee FID u soo koobaysaa qayb kasta oo ka mid ah sifooyinka sawirka ka hor inta aan la is barbar dhigin?
    • Sida hal celceliska qiimaha pixel
    • Sida liiska magacyada faylalka la soocay
    • Sida Gaussian badan oo kala duwan oo leh dhexdhexaadiye dhexdhexaad ah iyo matrix isku-dhafan
    • Sida histogaraam ee kanaalada midabka kaliya
  4. Waa maxay cilladda muhiimka ah ee Dhibcaha Bilowga ee FID ay wax ka qabatay?
    • Dhibcaha bilowga aad ayay u gaabis ahayd in la xisaabiyo
    • Dhibcaha bilowga waxay ku shaqaysay qoraalka oo kaliya
    • Dhibcaha bilowga uma baahna shabakad neerfaha
    • Dhibcaha bilowga weligood lama barbar dhigin sawirada la soo saaray iyo xogta dhabta ah
  5. When a generator produces less varied Inception features than the real images, how can that affect FID?
    • A covariance mismatch can contribute to a higher FID
    • FID must become zero
    • FID ignores feature diversity
    • FID automatically identifies the exact cause of the collapse
  6. In the Diffusion Models Beat GANs comparison, how was FID evaluated for ImageNet?
    • With 500 generated samples
    • With 50,000 generated samples
    • With 5,000 generated samples
    • With 5 million generated samples
  7. Why can a small evaluation set make an FID estimate unreliable?
    • FID requires every image to have a human label
    • The Inception network stops extracting features on small sets
    • Finite-sample bias can distort the estimated distance between feature distributions
    • Small sets force all covariance values to zero
  8. What does Kernel Inception Distance (KID) offer as an alternative in the guide?
    • A method for training the Inception-v3 network
    • A way to generate image captions
    • A measure of video frame rate
    • A metric used for evaluation when sample counts are smaller