Guide-yakabatana mibvunzo · Pakati Nhanho

Focal Loss for Imbalanced Detection Quiz

Inotarisa iyo yakatarisana-kurasikirwa modulator, kusaenzana kwekirasi, basa realpha negamma, uye kuti yakanyatso kurongeka uye yakaoma sei mienzaniso yakasiyana.

Nzira dzekutungamira dzinoenderanaFocal Kurasikirwa Kwekusagadzikana Kuonekwa
Mubvunzo 1 ye 8

Nderipi dambudziko guru muchikamu chimwe chekuonekwa kwechinhu chaive chekurasika kwakagadzirirwa kugadzirisa?

Tarisa mubvunzo wese8 mibvunzo

Mhinduro uye tsananguro zvinoonekwa paunenge uchitora bvunzo.

  1. Nderipi dambudziko guru muchikamu chimwe chekuonekwa kwechinhu chaive chekurasika kwakagadzirirwa kugadzirisa?
    • Kunyanya kusaenzana pakati pezvisingawanzo zvinhu zvemberi uye akawanda ari nyore kumashure
    • Inononoka inference kumhanya kwematanho maviri ekuona
    • Kushaikwa kwedata rekudzidziswa rakanyorwa
    • Overfitting kune zvinhu zvikuru
  2. Ndeipi iyo modulating factor iyo focal kurasikirwa inowanda pacross-entropy?
    • log(p_t)
    • (1 - p_t)^gamma
    • p_t^2 chete
    • 1 / p_t
  3. Chii chinoitika pakurasikirwa kwekutarisa kana iyo inotarisisa parameter gamma yakaiswa ku0?
    • Zvinova kureva kukanganisa kwakapetwa kaviri
    • Inofuratira mienzaniso yese yakanaka
    • Inova yakajairika cross-entropy
    • Inosiyana kusvika kukusingaperi
  4. Chii chinoita kuwedzera gamma kune mienzaniso iri nyore, yakanyatsorongedzwa?
    • Inowedzera mupiro wavo wekurasikirwa
    • Vanosiya mupiro wavo usina kuchinja
    • Inoabvisa kubva pane dataset
    • Inodzvanya mupiro wavo wekurasikirwa zvakanyanya
  5. Ndeipi detector architecture yakaunza focal kurasikirwa?
    • RetinaNet
    • YOLOv1
    • Inokurumidza R-CNN
    • AlexNet
  6. Ibasa rei iro alpha parameter inowanzoita padivi pegamma?
    • Inoisa mwero wekudzidza
    • Inoenzanisa huremu huremu hwemakirasi akanaka kana asina kunaka
    • Inodzora mufananidzo kugadzirisa
    • Inotsanangura nhamba yemabhokisi anchor
  7. For a hard example with p_t close to zero, what is the focal modulating factor (1 − p_t)^gamma likely to do?
    • Drive the factor to zero
    • Turn the example into a positive label
    • Keep the factor relatively close to one so the example retains substantial loss weight
    • Replace the cross-entropy term with alpha
  8. A dense detector has many easy background locations. What effect does focal loss aim to have on their combined training influence?
    • Increase their count by duplicating them
    • Remove all positive object examples
    • Make every example receive an identical loss
    • Reduce their dominance so training can focus more on sparse hard examples