An introduction to support Vector Machines: and other kernel-based learning methods
An introduction to support Vector Machines: and other kernel-based learning methods
Large-Scale Concept Ontology for Multimedia
IEEE MultiMedia
Naked image detection based on adaptive and extensible skin color model
Pattern Recognition
An adult image identification system employing image retrieval technique
Pattern Recognition Letters
Visual categorization with negative examples for free
MM '09 Proceedings of the 17th ACM international conference on Multimedia
Malicious content filtering based on semantic features
Proceedings of the 2nd International Conference on Interaction Sciences: Information Technology, Culture and Human
Evaluating feature selection for SVMs in high dimensions
ECML'06 Proceedings of the 17th European conference on Machine Learning
Neural network based adult image classification
ICANN'05 Proceedings of the 15th international conference on Artificial Neural Networks: biological Inspirations - Volume Part I
Semantic Home Photo Categorization
IEEE Transactions on Circuits and Systems for Video Technology
Hi-index | 0.00 |
Recently, in the Web and online social networking sites, the classification and filtering for naked images have been receiving a significant amount of attention. In our previous work, semantic feature in the aforementioned application is found to be more useful compared to using only low-level visual feature. In this paper, we further investigate the effective training strategy when making use of Support Vector Machine (SVM) for the purpose of generating semantic concept detectors. The proposed training strategy aims at increasing the performances of semantic concept detectors by boosting the 'naked' image classification performance. Extensive and comparative experiments have been carried out to access the effectiveness of proposed training strategy. In our experiments, each of the semantic concept detectors is trained with 600 images and tested with 300 images. In addition, 3 data sets comprising of 600 training images and 1000 testing images are used to test the naked image classification performance. The experimental results show that the proposed training strategy allows for improving semantic concept detection performance compared to conventional training strategy in use of SVM. In addition, by using our training strategy, one can improve the overall naked image classification performance when making use of semantic features.