evaluation

Updates on Performance of DMENet

All results are measured in matlab.

• Accuracy (Figure 2 in the main paper)

  • 0.8783

• Table 2 in the main paper (the last column)

  	...	Ours
  MSE	...	0.009
  MAE	...	0.077

  • We compute standard deviation maps (i.e., (out * 15) - 1) / 2) and clipped them to have values between 0 and 3.275 (evaluation/RTF/quantitative_RTF.m).

• Table 4 in the supplementary material (the last column)

  datasets	...	DMENetBDCS
  SYNDOF	...	0.013 / 0.084
  RTF	...	0.009 / 0.077

• Table 5 in the supplementary material (the last row)

  	SYNDOF MSE / MAE	RTF MSE / MAE	CUHK acc / mAP
  ...	...	...	...
  DMENetBDCS	0.013 / 0.084	0.009 / 0.077	0.878 / 0.987

• Table 6 in the supplementary material (the last column)

  Image #	...	Ours
  01	...	0.0643
  02	...	0.0406
  03	...	0.0863
  04	...	0.0408
  05	...	0.0335
  06	...	0.0756
  07	...	0.1129
  08	...	0.1695
  09	...	0.0427
  10	...	0.0771
  11	...	0.044
  12	...	0.1347
  13	...	0.0817
  14	...	0.1123
  15	...	0.0838
  16	...	0.0881
  17	...	0.0675
  18	...	0.0786
  19	...	0.0744
  20	...	0.0841
  21	...	0.0397
  22	...	0.0535
  Avg. MSE	...	0.0093
  Avg. MAE	...	0.0767
  Avg. MSE (s=1.0)	...	0.0207
  Avg. MAE (s=1.0)	...	0.1006
  Avg. MSE (s=1.6)	...	0.0491
  Avg. MAE (s=1.6)	...	0.1579