An Improved Convolutional Neural Network Algorithm and Its Application in Multilabel Image Labeling

Table 1

Comparison of the accuracy of labeling for each category in the Pascal VOC 2007 and Pascal VOC 2012 datasets based on different algorithms.

Image category	Labeling accuracy
	Zuo et al. [26]		Zhang et al. [27]		Islam et al. [28]		CNN		DCCNN
	2007	2012	2007	2012	2007	2012	2007	2012	2007	2012
Plane	0.802	0.777	0.988	0.973	0.947	0.924	0.992	0.983	1.0	0.999
Bike	0.501	0.425	0.812	0.748	0.498	0.451	0.905	0.877	0.996	0.973
Bird	0.561	0.454	0.873	0.808	0.962	0.946	1.0	0.977	1.0	0.984
Boat	0.619	0.533	0.899	0.853	0.671	0.652	0.935	0.920	0.990	0.972
Bottle	0.28	0.24	0.691	0.608	0.791	0.758	0.895	0.879	0.924	0.919
Bus	0.784	0.722	0.931	0.899	0.966	0.951	0.976	0.971	0.986	0.980
Car	0.584	0.506	0.897	0.868	0.905	0.891	0.953	0.949	0.989	0.987
Cat	0.607	0.542	0.941	0.893	0.941	0.923	0.962	0.955	0.984	0.970
Chair	0.509	0.453	0.613	0.554	0.422	0.39	0.826	0.794	0.907	0.893
Cow	0.309	0.26	0.848	0.778	0.866	0.857	1.0	0.999	1.0	1.0
Dining table	0.398	0.366	0.829	0.751	0.749	0.704	0.843	0.825	0.899	0.885
Dog	0.507	0.426	0.885	0.83	0.895	0.886	0.918	0.905	0.992	0.971
Horse	0.441	0.389	0.916	0.875	0.912	0.894	0.929	0.927	0.980	0.978
Motorbike	0.57	0.507	0.825	0.792	0.798	0.761	0.851	0.849	0.931	0.931
Person	0.769	0.703	0.899	0.847	0.831	0.794	0.899	0.897	0.962	0.957
Potted plant	0.305	0.234	0.636	0.578	0.681	0.658	0.833	0.829	0.884	0.881
Sheep	0.406	0.362	0.824	0.792	0.895	0.862	0.960	0.960	0.996	0.993
Sofa	0.385	0.314	0.449	0.395	0.389	0.339	0.725	0.719	0.819	0.816
Train	0.699	0.616	0.935	0.906	0.864	0.818	1.0	0.989	1.0	1.0
TV monitor	0.523	0.43	0.793	0.774	0.785	0.729	0.819	0.816	0.857	0.856
MAP value	0.528	0.463	0.824	0.776	0.788	0.759	0.911	0.901	0.955	0.947