Guide-yakabatana mibvunzo · Zvakaoma Nhanho

DETR Transformer Detection Quiz

Yedza DETR chinhu mibvunzo, Hungarian kuenzanisa, hapana-chinhu chakanangana, kupatsanura uye kurasikirwa kwebhokisi, uye chikonzero iyo set-based detector inodzivirira NMS.

Nzira dzekutungamira dzinoenderanaDetr Transformer Detection
Mubvunzo 1 ye 8

DETR inobvisa chii kana ichienzaniswa neyechinyakare ma detectors seFaster R-CNN?

Tarisa mubvunzo wese8 mibvunzo

Mhinduro uye tsananguro zvinoonekwa paunenge uchitora bvunzo.

  1. DETR inobvisa chii kana ichienzaniswa neyechinyakare ma detectors seFaster R-CNN?
    • Zvigadzirwa zvakagadzirwa nemaoko zvakadai semabhokisi anchor uye kwete-yakanyanya kudzvinyirira
    • Iyo CNN musana zvachose
    • Kudiwa kwechero data yekudzidziswa
    • Kusungirira mabhokisi semhando yekubuda
  2. Ndeipi algorithm inoshandiswa neDETR kuenzanisa fungidziro kune pasi-chokwadi zvinhu panguva yekudzidziswa?
    • k-zvinoreva kubatanidza
    • Gradient kuwedzera
    • Iyo Hungarian (bipartite) inoenderana algorithm
    • Beam search
  3. Chii chinonzi DETR's 'object queries'?
    • Mameseji ekurudziro anotaipa nemushandisi
    • Anchor mabhokisi ehukuru hwakagadziriswa
    • Pixels akatorwa kubva pamufananidzo
    • Seti yakamisikidzwa yekumisikidza kwakadzidzwa iyo decoder inoshanduka kuita yekuonekwa
  4. Ndechipi chivakwa chese chinosanganiswa neDETR?
    • Inodzokororwa network pamwe nemuti wesarudzo
    • A CNN musana pamwe neshanduro encoder-decoder
    • A yakachena GAN
    • A k-pedyo-muvakidzani classifier
  5. How does DETR's training objective support inference without a separate NMS stage?
    • They make every possible duplicate prediction mathematically impossible
    • The model can only output one object per image
    • One-to-one matching and set losses discourage multiple predictions for the same object
    • A hidden NMS classifier supplies every training target
  6. Ndechipi chaive kushaya simba kwaizivikanwa kweDETR yekutanga?
    • Kunonoka kudzidzisa kusangana uye kuona kushoma-chinhu chidiki
    • Yakatadza kumhanya paGPUs
    • Zvaida hapana data yakanyorwa
    • Yakangoshanda pamifananidzo yegreyscale
  7. What training label does original DETR assign to an object query that is not matched to a ground-truth object?
    • The largest visible object's class regardless of matching
    • The no-object class
    • A mandatory new ground-truth object invented for the query
    • The image's filename as a class label
  8. Which prediction errors does DETR's set loss penalize on matched object pairs?
    • Only the number of letters in the class name
    • Only the image's compression size
    • Only whether the training batch uses a GPU
    • Both the predicted class and bounding-box errors