TsCNNs-Based Inappropriate Image and Video Detection System for a Social Network
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
We propose a detection algorithm based on tree-structured convolutional neural networks (TsCNNs) that findspornography, propaganda, or other inappropriate content on a social media network. The algorithm sequentiallyapplies the typical convolutional neural network (CNN) algorithm in a tree-like structure to minimizeclassification errors in similar classes, and thus improves accuracy. We implemented the detection system andconducted experiments on a data set comprised of 6 ordinary classes and 11 inappropriate classes collected fromthe Korean military social network. Each model of the proposed algorithm was trained, and the performancewas then evaluated according to the images and videos identified. Experimental results with 20,005 new imagesshowed that the overall accuracy in image identification achieved a high-performance level of 99.51%, and theeffectiveness of the algorithm reduced identification errors by the typical CNN algorithm by 64.87 %. Byreducing false alarms in video identification from the domain, the TsCNNs achieved optimal performance of98.11% when using 10 minutes frame-sampling intervals. This indicates that classification through propersampling contributes to the reduction of computational burden and false alarms.
Abstract
We propose a detection algorithm based on tree-structured convolutional neural networks (TsCNNs) that findspornography, propaganda, or other inappropriate content on a social media network. The algorithm sequentiallyapplies the typical convolutional neural network (CNN) algorithm in a tree-like structure to minimizeclassification errors in similar classes, and thus improves accuracy. We implemented the detection system andconducted experiments on a data set comprised of 6 ordinary classes and 11 inappropriate classes collected fromthe Korean military social network. Each model of the proposed algorithm was trained, and the performancewas then evaluated according to the images and videos identified. Experimental results with 20,005 new imagesshowed that the overall accuracy in image identification achieved a high-performance level of 99.51%, and theeffectiveness of the algorithm reduced identification errors by the typical CNN algorithm by 64.87 %. Byreducing false alarms in video identification from the domain, the TsCNNs achieved optimal performance of98.11% when using 10 minutes frame-sampling intervals. This indicates that classification through propersampling contributes to the reduction of computational burden and false alarms.
참고문헌 (16)
- S. Sharma, "Proliferation of social computing : cultural computing paradigm", International Journal of Computer Applications, 137(9), pp.27~30 (2016)
- "Daily time spent on social networking by internet users worldwide from 2012 to 2022" (2022)[URL]
- Y. C. Lin, "Pornography detection using support vector machine", Proceedings of the 16th IPPR Conference on Computer Vision, Graphics and Image Processing (CVGIP), pp.123~130 (2003)
- A. Krizhevsky, "ImageNet classification with deep convolutional neural networks", Advances in Neural Information Processing Systems, 25, pp.1106~1114 (2012)
- M. M. Fleck, "Finding naked people", Computer Vision – ECCV ‘96, pp.593~602 (1996)
- C. Y. Jeong, "Appearance-based nude image detection", Proceedings of the 17th International Conference on Pattern Recognition (ICPR), pp.467~470 (2004)
- Q. F. Zheng, "Shape-based adult image detection", International Journal of Image and Graphics, 6(1), pp.115~124 (2006)
- H. A. Rowley, "Large scale image-based adult-content filtering", Proceedings of the First International Conference on Computer Vision Theory and Applications (VISAPP), pp.290~296 (2006)
- T. Deselaers, "Bag-of-visual-words models for adult image classification and filtering", Proceedings of 2008 19th International Conference on Pattern Recognition, pp.1~4 (2008)
- S. Avila, "Pooling in image representation : the visual codeword point of view", Computer Vision and Image Understanding, 117(5), pp.453~465 (2013)
- C. Caetano, "Pornography detection using BossaNova video descriptor", Proceedings of 2014 22nd European Signal Processing Conference (EUSIPCO), pp.1681~1685 (2014)
- "Very deep convolutional networks for large-scale image recognition" (2014)[URL]
- C. Szegedy, "Going deeper with convolutions", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.1~9 (2015)
- K. He, "Deep residual learning for image recognition", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.770~778 (2016)
- F. Nian, "Pornographic image detection utilizing deep convolutional neural networks", Neurocomputing, 210, pp.283~293 (2016)
- M. Abadi, "TensorFlow: a system for large-scale machine learning", Proceedings of the 12th USENIX Symposium on Operating Systems Design and Implementation (OSDI), pp.265~283 (2016)