TABLE II.
Average Classification Rate for Each Decision Fusion Approach Majority Vote Rule and Maximum Confidence Score for Both KNN and SVM Classifiers for Multiple Candidates (1, and 8)
| Decision Fusion (Classifier) |
1 Candidate |
8 Candidates |
|---|---|---|
|
Majority vote rule (KNN) |
0.70 | 0.74 |
|
Maximum confidence score (KNN) |
<0.1 | 0.75 |
|
Majority vote rule (SVM) |
0.57 | 0.72 |
|
Maximum confidence score (SVM) |
0.33 | 0.70 |
| SVM concat. | 0.52 | |
| SIFT FV | 0.61 | |
And comparison with SVM feature concatenation (SVM concat.) and Fisher vector encoding with SIFT (SIFT FV).