Skip to main content

Table 3 Classification performance on the ICPR 2012 dataset

From: Automated classification for HEp-2 cells based on linear local distance coding framework

  Cell level Cell level Image level
Algorithm Classification Recall (%) Classification
  accuracy (%)   accuracy (%)
LSC-(sift+ldv)a 71.7 72.9 85.7
LSC-(sift+dv)b 67.7 69.2 78.6
LSC-ldvc 67.4 68.9 78.6
LSC-siftd 66.4 68.1 78.6
LLC-(sift+ldv)e 70.9 71.6 78.6
LLC-(sift+dv)f 69.3 70.6 78.6
LLC-ldvg 69.1 70.9 78.6
LLC-sifth 68.9 70.1 78.6
  1. The italicized items represent the best performance.
  2. aThe LLDC method based on the concatenated representations from LSC-sift and LSC-ldv.
  3. bThe LDC method based on the concatenated representations from LSC-sift and LSC-dv.
  4. cThe LSC method based on local distance vector.
  5. dThe LSC method based on SIFT features.
  6. eThe LLDC method based on the concatenated representations from LLC-sift and LLC-ldv.
  7. fThe LDC method based on the concatenated representations from LLC-sift and LLC-dv.
  8. gThe LLC method based on local distance vector.
  9. hThe LLC method based on SIFT features.