그림 1. CNN 구조 Fig. 1. CNN Architecture
그림 2. 제안 MNN 구조 Fig. 2. Prosed MNN architecture
그림 3. MNN 커널 Fig. 3. MNN Kernel
그림 4. 테스트 영상(daisy,dandelion,rose,tulips) Fig. 4. Test image(daisy,dandelion,rose,tulips)
그림 5. 에로전과 다이레이션 영상 Fig. 5. Erosion and Dilation image
그림 6. 커널에 따른 특징 차이 Fig. 6. Feature difference by kernel
그림 7. Pooling-ReLU 영상 Fig. 7. Pooling-ReLU image
그림 8. Fully connected 레이어 Fig. 8. Fully connected layer
그림 9. 구글 인셉션 구조 Fig. 9. Google Inception Architecture
표 1, CNN(구글 인셉션 모듈)과 MNN의 인식 결과 Table 1. Recognition results in CNN(Google inception module) and MNN
References
- Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton, "ImageNet Classification with Deep Convolutional Neural Networks", Advances in Neural Information processing Systems 25, NIPS, 2012.
- H. Lee, R. Grosse, R. Ranganath, and A.Y. Ng., "Convolutional deep belief networks for scalable unsupervised learning of hierarchical represent ations", Proceedings of the 26th Annual International Conference on Machine Learning, ACM, 2009.
- Y. LeCun, K. Kavukcuoglu, and C. Farabet, "Convolutional networks and applications in vision", International Symposium in Circuits and Systems (ISCAS), IEEE, 2010.
- Serra,J., Image Analysis and Mathe_ matical Morphology, Vol.1, Academic Press, New York, 1982.
- Serra, J., "Introduction to Mathematical Morphology," Computer Vision, Graphics, and Image Processing, Vol.35, No.3, 1986.
- Jongho, Choi and Jongsoo, Choi, "A Morphological Recognition Algorithm of Blob Shaped Particles for Automated Visual Inspection", Journal of KIISE, Vol. 21, No.5, May, 1994.
- Jong-Ho Choi, "Morphological Object Recognition Algorithm", Journal of KIIECT, Vol. 11, No. 2, April, 2018.
- Szegedy, Christian,et al. "Going deeper with convolutions." Proceedings of the IEEE conference on computer vision and pattern recognition. 2015.