MODEL SELECTION AND HYPERPARAMETER OPTIMIZATION FOR FLOWER CLASSIFICATION USING OPEN-SOURCE IMAGE DATA

International Journal of Computer Science (IJCS) Published by SK Research Group of Companies (SKRGC)

Abstract

Image classification plays a critical role in computer vision, enabling the categorization of images into predefined classes. With advancements in machine learning—particularly deep learning—image classification has seen significant improvements in both accuracy and efficiency. This study focuses on flower image classification using the Oxford 102 Flower Dataset, which includes images across 102 flower categories. The primary aim is to compare various machine learning models and kernel functions to optimize classification performance. The research examines different kernel types used with Support Vector Machines (SVM), including linear, polynomial, RBF, and sigmoid, while also exploring model optimization techniques. A Convolutional Neural Network (CNN)-based model is proposed and evaluated against established architectures like LeNet, AlexNet, and VggNet. The model uses ReLU for feature learning and Softmax for classification. Performance is assessed using metrics such as accuracy, precision, recall, and F1-score. Results demonstrate that the proposed model not only outperforms traditional CNNs in classification accuracy but also shows faster convergence and improved generalization due to effective optimization and hyper parameter tuning. Overall, the findings highlight the potential of deep learning, particularly optimized CNNs, in handling complex image classification tasks. The study underscores the importance of model architecture and tuning strategies in achieving robust and scalable solutions for real-world applications.

References

1. M.-E. Nilsback and A. Zisserman, “Automated flower classification over a large number of classes,” in Proceedings of the Indian Conference on Computer Vision, Graphics and Image Processing, Dec. 2008, pp. 722–729.

2. L. Y. Zhang, X. Y. Wang, and D. X. Wu, "Image Classification Using Deep Learning," International Journal of Computer Vision, vol. 6, pp. 1-13, 2017.

3. A. Y. Chen and M. L. Zhong, "Deep Convolutional Networks for Image Classification: A Review," IEEE Transactions on Image Processing, vol. 23, no. 12, pp. 5121-5132, Dec. 2018.

4. D. J. Kingma and J. Ba, "Adam: A Method for Stochastic Optimization," International Conference on Learning Representations (ICLR), 2015.

5. Y. He, Y. Song, and Z. Zhang, "Softmax Function and Its Application in Deep Neural Networks," IEEE Transactions on Neural Networks, vol. 29, no. 3, pp. 2345-2356, March 2019.

6. Sermanet, P., Eigen, D., & Zhang, X. (2014). "OverFeat: Integrated recognition, localization and detection using convolutional networks." International Conference on Learning Representations (ICLR).

7. Nilsback, M. E., & Zisserman, A. (2012). "A visual vocabulary for flower classification." IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 1447-1454.

8. Simonyan, K., & Zisserman, A. (2014). "Very deep convolutional networks for large-scale image recognition." International Conference on Learning Representations (ICLR).

9. He, K., Zhang, X., Ren, S., & Sun, J. (2015). "Deep residual learning for image recognition." IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 770-778.

10. LeCun, Y., Bengio, Y., & Hinton, G. (2015). "Deep learning." Nature, 521(7553), 436-444.

11. Gao, W., Li, Q., & Li, H. (2016). "Comparison of kernel methods for image classification using the Oxford 102 Flower dataset." International Journal of Computer Vision, 109(1), 34-45.

12. Zhang, X., & Song, H. (2016). "Transfer learning for image classification with pre-trained convolutional neural networks on Oxford 102 flower dataset." Neural Networks, 76, 106-113.

13. Karpathy, A., Hejrati, M., & Fei-Fei, L. (2014). "Deep visual-semantic alignments for generating image descriptions." IEEE Transactions on Pattern Analysis and Machine Intelligence, 36(9), 1688-1700.

14. Dosovitskiy, A., & Fischer, P. (2014). "Discriminative Unsupervised Feature Learning with Exemplar Convolutional Neural Networks." IEEE Transactions on Pattern Analysis and Machine Intelligence, 38(9), 1734-1747.

15. Bertasius, G., & Shi, J. (2016). "Learning category-specific features for fine-grained image classification." IEEE Transactions on Neural Networks and Learning Systems, 27(8), 1740-1752.

16. Huang, G., Liu, Z., & Van Der Maaten, L. (2017). "Densely connected convolutional networks." IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 4700-4708.

17. Szegedy, C., Vanhoucke, V., & Ioffe, S. (2016). "Rethinking the Inception Architecture for Computer Vision." IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2818-2826.

18. Yuan, L., & Fu, L. (2015). "Multi-feature representation for image classification using Oxford 102 Flower dataset." Journal of Visual Communication and Image Representation, 28, 81-90.

19. Zhou, B., & Chen, L. (2018). "Optimizing feature representations for flower image classification." Journal of Computer Vision and Applications, 30(2), 112-122.

20. Tan, C., Sun, F., & Kong, T. (2016). "Comparative study of deep learning models on Oxford 102 Flower dataset." International Journal of Computer Vision, 120(5), 274-290.

21. K. Simonyan and A. Zisserman, "Very deep convolutional networks for large-scale image recognition," International Conference on Learning Representations (ICLR), 2014.

22. K. He, X. Zhang, S. Ren, and J. Sun, "Deep residual learning for image recognition," IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2015. All Rights Reserved ©2019 International Journal of Computer Science (IJCS Journal) Published by SK Research Group of Companies (SKRGC) - Scholarly Peer Reviewed Research Journals http://www.skrgcpublication.org/ Page 2381

23. C. Szegedy, V. Vanhoucke, S. Ioffe, J. Shlens, and Z. Wojna, "Rethinking the Inception Architecture for Computer Vision," IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016.

24. C. Szegedy, W. Liu, Y. Jia, P. Sermanet, and S. Reed, "Going deeper with convolutions," IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2014.

25. A. Krizhevsky, I. Sutskever, and G. E. Hinton, "ImageNet classification with deep convolutional neural networks," Advances in Neural Information Processing Systems (NIPS), 2012. For the reference used in the above text, you can cite the following source:

26. A. Krizhevsky, I. Sutskever, and G. E. Hinton, "ImageNet classification with deep convolutional neural networks," Communications of the ACM, vol. 60, no. 6, pp. 84-90, Jun. 2017.

27. A. Graves, "Supervised sequence labelling with recurrent neural networks," in Lecture Notes in Computer Science, vol. 3858, Springer, 2006, pp. 58-85.

Keywords

Image Classification, Deep Learning, Support Vector Machines (SVM), Convolutional Neural Networks (CNN), Oxford 102 Flower Dataset, Model Optimization.