Preprint / Version 1

ANALISIS DAMPAK TRANSFER LEARNING PADA SEGMENTASI SEMANTIK CITRA HEWAN MENGGUNAKAN U-NET

Keywords:

convolutional neural network, deep learning, pengolahan citra, segmentasi sematik, transfer learning, u-net

Abstract

Segmentasi merupakan salah satu metode pengolahan citra yang membagi citra menjadi beberapa segmen. Segmentasi pada percobaan ini menggunakan metode deep learning, yaitu convolutional neural network (CNN) dengan arsitektur U-Net. CNN adalah salah satu metode deep learning yang sangat populer pada pengolahan citra dan arsitektur U-Net biasa digunakan untuk segmentasi citra berjenis semantik. Hasil segmentasi dianalisa menggunakan metrik Intersection-over-Union (IoU) menunjukkan bahwa model U-Net dengan transfer learning memiliki performa yang lebih baik dibandingkan dengan model U-Net tanpa transfer learning.

References

Guo, D., Pei, Y., Zheng, K., Yu, H., Lu, Y., & Wang, S. (2020). Degraded Image Semantic Segmentation with Dense-Gram Networks. IEEE Transactions on Image Processing, 29, 782–795. https://doi.org/10.1109/TIP.2019.2936111

Intersection over Union (IoU) – Johannes S. Fischer. (n.d.). Retrieved December 27, 2021, from https://johfischer.com/2021/11/04/intersection-over-union-iou/

Visual Geometry Group - University of Oxford. (n.d.). Retrieved December 27, 2021, from https://www.robots.ox.ac.uk/~vgg/data/pets/

Image Augmentation. Improving Deep learning models | by Sanchit Tanwar | Analytics Vidhya | Medium. (n.d.). Retrieved December 27, 2021, from https://medium.com/analytics-vidhya/image-augmentation-9b7be3972e27

Ronneberger, O., Fischer, P., & Brox, T. (n.d.). U-Net: Convolutional Networks for Biomedical Image Segmentation. http://lmb.informatik.uni-freiburg.de/

Howard, A. G., Zhu, M., Chen, B., Kalenichenko, D., Wang, W., Weyand, T., & Andreetto, M. (n.d.). MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications.

Yudistira, Novanto, et al. "Prediction of sequential organelles localization under imbalance using a balanced deep u-net." Scientific reports 10.1 (2020): 1-11.

Hakim, Lukman, et al. "Regularizer based on Euler characteristic for retinal blood vessel segmentation." Pattern Recognition Letters 149 (2021): 83-90.

Hakim, Lukman, et al. "U-net with graph based smoothing regularizer for small vessel segmentation on fundus image." International Conference on Neural Information Processing. Springer, Cham, 2019.

Downloads

Posted

2022-06-07