Maswali yanayounganishwa na mwongozo · Kati Kiwango

Maswali ya Kuanzishwa kwa Fréchet

Kagua ulinganisho wa nafasi ya vipengele vya FID ya Gaussian na unyeti wake wa ukubwa wa sampuli, pamoja na njia mbadala za kutathmini uaminifu na utofauti wa picha zinazozalishwa.

Njia za mwongozo zinazohusianaUmbali wa Kuanzishwa kwa Fr Chet
Swali 1 ya 8

Alama ya chini ya FID inaonyesha nini?

Hakiki kila swali8 maswali

Majibu na maelezo huonekana unapojibu maswali.

  1. Alama ya chini ya FID inaonyesha nini?
    • Takwimu za picha zinazozalishwa ziko karibu na zile za picha halisi
    • Picha zinazozalishwa sio za kweli
    • Muundo huo ulifunzwa kwa enzi chache
    • Picha zina azimio la juu zaidi
  2. Je, ni mtandao gani uliofunzwa awali ambao FID ya kawaida hutumia kutoa vipengele vya picha?
    • ResNet-50
    • Kuanzishwa-v3
    • VGG-16
    • BERT
  3. FID inafanyaje muhtasari wa kila seti ya vipengee vya picha kabla ya kuvilinganisha?
    • Kama thamani moja ya wastani ya pikseli
    • Kama orodha iliyopangwa ya majina ya faili
    • Kama Gaussian ya aina nyingi iliyo na vekta ya maana na matrix ya ushirika
    • Kama histogram ya chaneli za rangi pekee
  4. Je, FID ilishughulikia upungufu gani mkuu wa Alama ya Kuanzishwa?
    • Alama ya Kuanzishwa ilikuwa polepole sana kukokotoa
    • Alama ya Kuanzishwa ilifanya kazi kwenye maandishi pekee
    • Alama ya Kuanzishwa haihitaji mtandao wa neva
    • Alama ya Kuanzishwa haijawahi kulinganisha picha zinazozalishwa na data halisi
  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