Skip to main content
. 2018 Apr 25;13(4):e0194526. doi: 10.1371/journal.pone.0194526

Table 1. Performance comparison and statistical analysis of different sizes of the dictionary and features percentages of the image on the Corel-1000 image collection (bold values indicate best performance).

Features % per image Performance analysis in terms of the MAP performance (in %) on the different sizes of the dictionary
20 50 100 200 400 600 800 1000 1200
10% 74.06 75.60 79.53 79.86 80.76 80.76 80.86 80.85 80.53
25% 76.46 78.33 81.16 80.93 81.16 80.99 81.26 81.19 81.25
50% 73.93 78.19 80.60 81.19 81.93 82.30 86.00 80.66 81.13
75% 74.40 77.93 80.40 82.05 81.73 82.05 82.06 81.00 81.33
100% 75.60 76.80 80.26 82.40 80.73 82.20 82.40 81.13 81.53
MAP 74.89 77.37 80.39 81.28 81.26 81.66 82.51 80.96 81.15
Std. dev. 1.097 1.158 0.590 0.999 0.550 0.726 2.042 0.215 0.377
Confidence interval 73.52-
76.25
75.93-
78.80
79.65-
81.12
80.04-
82.52
80.57-
81.94
80.75-
82.56
80.80-
83.10
80.69-
81.23
80.68-
81.62
Std. error 0.490 0.517 0.263 0.446 0.246 0.324 0.413 0.096 0.169
Non-parametric statistical analysis using Wilcoxon matched-pairs signed-rank test
Z-value 2.023 2.023 2.023 2.023 2.023 2.023 2.023 2.023 2.023
P-value 0.043 0.043 0.043 0.043 0.043 0.043 0.043 0.043 0.043