Pengenalan Varietas Ikan Koi Berdasarkan Foto Menggunakan Simple Linear Iterative Clustering Superpixel Segmentation dan Convolutional Neural
DOI:
https://doi.org/10.17977/um068v1i112021p806-814Kata Kunci:
superpixel, SLIC, graph, machine learning, neural networkAbstrak
Object segmentation and image recognition are two computer vision tasks which are still being developed until today. Simple Linear Iterative Clustering is an algorithm which is very popular to help with object segmentation tasks because it is the best in terms of result and speed. In image recognition, Convolutional Neural Networks are also one of the best approaches for any kind of recognition tasks because of their efficiency and the ability to recognize objects like animals do. Koi fish have become a very interesting object to be researched because they are difficult to segment and distinguished between their varieties. The dataset consists of 600 images of Koi fish from 10 different varieties. The Koi fish’s recognition process begins with generating super pixels for the input image. The next step is to merge all neighborhood super pixels by their color similarities. After this step, almost all the background pixels should be detected so that the actual object, the Koi fish, can be segmented. The segmented image is then given to a Convolutional Neural Networks, to learn any important features which distinguish every Koi fish variety from one another. A trained Convolutional Neural Networks can then give a Koi fish variety prediction for an input image. Based on a series of segmentation and model tests performed, it is proven that the segmentation technique, which uses Simple Linear Iterative Clustering in this project, performs exceptionally well across almost all the images in the dataset. The model produced from this project is also able to classify a wide range of Koi fish varieties accurately at 90 percent accuracy with segmentation and 87 percent without segmentation.
Segmentasi dan pengenalan objek pada gambar masih merupakan dua buah masalah pada computer vision yang masih terus diteliti dan dikembangkan hingga saat ini. Simple Linear Iterative Clustering merupakan salah satu algoritma segmentasi superpixel yang cukup populer untuk membantu melakukan segmentasi objek karena memiliki hasil superpixel yang baik dan dapat berjalan dengan cepat. Untuk pengenalan objek, Convolutional Neural Networks masih merupakan salah satu yang terbaik untuk berbagai masalah karena efisien dan mampu mengenali objek pada gambar layaknya hewan mengenali objek yang dilihatnya. Ikan koi menjadi sebuah objek yang menarik untuk diteliti karena sulit untuk disegmentasi dan dikenali jenisnya bahkan oleh manusia. Dataset yang digunakan berisi 600 gambar yang terdiri dari 10 varietas ikan koi. Pengenalan ikan koi diawali dengan melakukan generate superpixel pada gambar input, kemudian menggabungkan superpixel-superpixel terdekat yang memiliki warna yang mirip. Dengan cara ini, maka hampir seluruh pixel background dapat dideteksi sehingga objek ikan koi dapat disegmentasi. Gambar hasil segmentasi kemudian dilatihkan ke Convolutional Neural Networks yang akan mempelajari fitur-fitur penting pada setiap jenis ikan koi yang diteliti. Convolutional Neural Networks yang telah dilatih kemudian dapat memberikan prediksi varietas ikan koi dari sebuah input gambar. Berdasarkan hasil uji coba segmentasi dan model yang digunakan, dibuktikan bahwa teknik segmentasi yang memanfaatkan Simple Linear Iterative Clustering yang dilakukan berhasil untuk hampir seluruh gambar pada dataset. Model yang dibuat mampu mengklasifikasikan varietas ikan koi dengan akurasi 90 persen dengan segmentasi dan 87 persen tanpa segmentasi.
Referensi
Achanta, R., Shaji, A., Smith, K., Lucchi, A., Fua, P., & Süsstrunk, S. (2010). Slic superpixels (No. EPFL-REPORT-149300).
Achanta, R., Shaji, A., Smith, K., Lucchi, A., Fua, P., & Süsstrunk, S. (2012). SLIC superpixels compared to state-of-the-art superpixel methods. IEEE transactions on pattern analysis and machine intelligence, 34(11), 2274-2282.
He, S., Lau, R. W., Liu, W., Huang, Z., & Yang, Q. (2015). Supercnn: A superpixelwise convolutional neural network for salient object detection. International Journal of Computer Vision, 115(3), 330-344.
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017). Imagenet classification with deep convolutional neural networks. Communications of the ACM, 60(6), 84-90.
LeCun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998). Gradient-based learning applied to document recognition. Proceedings of the IEEE, 86(11), 2278-2324.
Limantoro, S. E., Kristian, Y., & Purwanto, D. D. (2017). Deteksi Pengendara Sepeda Motor Menggunakan Deep Convolutional Neural Networks. In Seminar Nasional Teknologi Informasi dan Komunikasi (pp. 79-86).
Purnomo, M. H., Kristian, Y., Setyati, E., Rosiani, U. D., & Setiawan, E. I. (2016, October). Limitless possibilities of pervasive biomedical engineering: Directing the implementation of affective computing on automatic health monitoring system. In 2016 8th International Conference on Information Technology and Electrical Engineering (ICITEE) (pp. 1-4). IEEE.
Rastegari, M., Ordonez, V., Redmon, J., & Farhadi, A. (2016, September). Xnor-net: Imagenet classification using binary convolutional neural networks. In Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part IV (pp. 525-542). Cham: Springer International Publishing.
Redmon, J., & Farhadi, A. (2017). YOLO9000: better, faster, stronger. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 7263-7271).
Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. (2016). You only look once: Unified, real-time object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 779-788)..
Shi, J., & Malik, J. (2000). Normalized cuts and image segmentation. IEEE Transactions on pattern analysis and machine intelligence, 22(8), 888-905.
Simonyan, K., & Zisserman, A. (2014). Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409. 1556.
Springenberg, J. T., Dosovitskiy, A., Brox, T., & Riedmiller, M. (2014). Striving for simplicity: All convolutional net. arXiv preprint arXiv:1412.6806.
Trémeau, A., & Colantoni, P. (2000). Region’s adjacency graph applied to color image segmentation. IEEE Transactions on image processing, 9(4), 735-744.
3.png)
1.png)
1.png)
