Person Detection for an Orthogonally Placed Monocular Camera

Table 3

Evaluation of classification performance of ConvNet-based (first section) and HOG-based (second section) passenger recognition systems using the measures (3)–(6).

Classifier	Accuracy	Precision	Recall	F1-score
Net1	0.949	0.950	0.948	0.949
Net2	0.953	0.947	0.961	0.954
LeNet-5	0.956	0.946	0.966	0.956
AlexNet	0.947	0.921	0.977	0.948
VGG_16	0.928	0.903	0.958	0.930
SVM with RBF kernel function, cell size [6, 6] px	0.949	0.957	0.941	0.949
SVM with linear kernel function, cell size [6, 6] px	0.939	0.947	0.931	0.939
SVM, polynomial degree = 2, cell size [6, 6] px	0.946	0.946	0.947	0.946
SVM, polynomial degree = 3, cell size [6, 6] px	0.953	0.953	0.947	0.950
SVM with RBF kernel function, cell size [8, 8] px	0.949	0.952	0.945	0.948
SVM with linear kernel function, cell size [8, 8] px	0.938	0.941	0.934	0.937
SVM, polynomial degree = 2, cell size [8, 8] px	0.947	0.951	0.943	0.947
SVM, polynomial degree = 3, cell size [8, 8] px	0.948	0.949	0.946	0.948
SVM with RBF kernel function, cell size [10, 10] px	0.956	0.964	0.947	0.956
SVM with linear kernel function, cell size [10, 10] px	0.943	0.946	0.939	0.943
SVM, polynomial degree = 2, cell size [10, 10] px	0.947	0.947	0.948	0.948
SVM, polynomial degree = 3, cell size [10, 10] px	0.959	0.957	0.961	0.959
SVM with RBF kernel function, cell size [12, 12] px	0.953	0.956	0.948	0.952
SVM with linear kernel function, cell size [12, 12] px	0.935	0.939	0.930	0.934
SVM, polynomial degree = 2, cell size [12, 12] px	0.950	0.955	0.945	0.950
SVM, polynomial degree = 3, cell size [12, 12] px	0.950	0.957	0.942	0.949
SVM with RBF kernel function, cell size [14, 14] px	0.929	0.925	0.913	0.919
SVM with linear kernel function, cell size [14, 14] px	0.919	0.925	0.913	0.919
SVM, polynomial degree = 2, cell size [14, 14] px	0.929	0.936	0.922	0.929
SVM, polynomial degree = 3, cell size [14, 14] px	0.921	0.920	0.923	0.922
SVM with RBF kernel function, cell size [16, 16] px	0.952	0.955	0.949	0.952
SVM with linear kernel function, cell size [16, 16] px	0.942	0.949	0.934	0.941
SVM, polynomial degree = 2, cell size [16, 16] px	0.948	0.951	0.944	0.948
SVM, polynomial degree = 3, cell size [16, 16] px	0.943	0.943	0.943	0.943

Note: best results are in bold.