• The Journal of the Acoustical Society of Korea
• /
• v.41 no.3
• /
• pp.351-358
• /
• 2022

Recently, Generative Adversarial Networks (GAN) and Variational AutoEncoders (VAE) have been applied to voice conversion that can make use of non-parallel training data. Especially, Conditional Cycle-Consistent Generative Adversarial Networks (CC-GAN) and Cycle-Consistent Variational AutoEncoders (CycleVAE) show promising results in many-to-many voice conversion among multiple speakers. However, the number of speakers has been relatively small in the conventional voice conversion studies using the CC-GANs and the CycleVAEs. In this paper, we extend the number of speakers to 100, and analyze the performances of the many-to-many voice conversion methods experimentally. It has been found through the experiments that the CC-GAN shows 4.5 % less Mel-Cepstral Distortion (MCD) for a small number of speakers, whereas the CycleVAE shows 12.7 % less MCD in a limited training time for a large number of speakers.

• The Journal of the Acoustical Society of Korea
• /
• v.39 no.4
• /
• pp.292-302
• /
• 2020

Machine learning algorithms have made immense contributions in various fields including sonar and radar applications. Recently developed Cycle-Consistency Generative Adversarial Network (CycleGAN), a variant of GAN has been successfully used for unpaired image-to-image translation. We present a modified CycleGAN for translation of underwater ship engine sounds with high perceptual quality. The proposed network is composed of an improved generator model trained to translate underwater audio from one vessel type to other, an improved discriminator to identify the data as real or fake and a modified cycle-consistency loss function. The quantitative and qualitative analysis of the proposed CycleGAN are performed on publicly available underwater dataset ShipsEar by evaluating and comparing Mel-cepstral distortion, pitch contour matching, nearest neighbor comparison and mean opinion score with existing algorithms. The analysis results of the proposed network demonstrate the effectiveness of the proposed network.

• Korean Journal of Remote Sensing
• /
• v.38 no.1
• /
• pp.33-43
• /
• 2022

This study presents a method to restore an optical satellite image with distortion and occlusion due to fog, haze, and clouds to one that minimizes degradation factors by referring to the same type of peripheral image. Specifically, the time and cost of re-photographing were reduced by partially occluding a region. To maintain the original image's pixel value as much as possible and to maintain restored and unrestored area continuity, a simulation restoration technique modified with the Cycle Generative Adversarial Network (CycleGAN) method was developed. The accuracy of the simulated image was analyzed by comparing CycleGAN and histogram matching, as well as the pixel value distribution, with the original image. The results show that for Site 1 (out of three sites), the root mean square error and R² of CycleGAN were 169.36 and 0.9917, respectively, showing lower errors than those for histogram matching (170.43 and 0.9896, respectively). Further, comparison of the mean and standard deviation values of images simulated by CycleGAN and histogram matching with the ground truth pixel values confirmed the CycleGAN methodology as being closer to the ground truth value. Even for the histogram distribution of the simulated images, CycleGAN was closer to the ground truth than histogram matching.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.38 no.6
• /
• pp.499-509
• /
• 2020

This study explores how to build object classification learning data based on artificial intelligence. The data has been investigated recently in image classification fields and, in turn, has a great potential to use. In order to recognize and extract relatively accurate objects using artificial intelligence, a large amount of learning data is required to be used in artificial intelligence algorithms. However, currently, there are not enough datasets for object recognition learning to share and utilize. In addition, generating data requires long hours of work, high expenses and labor. Therefore, in the present study, a small amount of initial aerial image learning data was used in the GAN (Generative Adversarial Network)-based generator network in order to establish image learning data. Moreover, the experiment also evaluated its quality in order to utilize additional learning datasets. The method of oversampling learning data using GAN can complement the amount of learning data, which have a crucial influence on deep learning data. As a result, this method is expected to be effective particularly with insufficient initial datasets.

• Information Systems Review
• /
• v.24 no.4
• /
• pp.97-118
• /
• 2022

As generative adversarial network (GAN) based oversampling techniques have achieved impressive results in class imbalance of unstructured dataset such as image, many studies have begun to apply it to solving the problem of imbalance in structured dataset. However, these studies have failed to reflect the characteristics of structured data due to changing the data structure into an unstructured data format. In order to overcome the limitation, this study adapted CycleGAN to reflect the characteristics of structured data, and proposed hybridization of synthetic minority oversampling technique (SMOTE) and the adapted CycleGAN. In particular, this study tried to overcome the limitations of existing studies by using a one-dimensional convolutional neural network unlike previous studies that used two-dimensional convolutional neural network. Oversampling based on the method proposed have been experimented using various datasets and compared the performance of the method with existing oversampling methods such as SMOTE and adaptive synthetic sampling (ADASYN). The results indicated the proposed hybrid oversampling method showed superior performance compared to the existing methods when data have more dimensions or higher degree of imbalance. This study implied that the classification performance of oversampling structured data can be improved using the proposed hybrid oversampling method that considers the characteristic of structured data.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.40 no.3
• /
• pp.177-185
• /
• 2022

A method of restoring the occluded area was proposed by referring to images taken with the same types of sensors on high-resolution optical satellite images through deep learning. For the natural continuity of the simulated image with the occlusion region and the surrounding image while maintaining the pixel distribution of the original image as much as possible in the patch segmentation image, CycleGAN (Cycle Generative Adversarial Network) method with ConvNeXt block applied was used to analyze three experimental regions. In addition, We compared the experimental results of a training patch size of 512*512 pixels and a 1024*1024 pixel size that was doubled. As a result of experimenting with three regions with different characteristics,the ConvNeXt CycleGAN methodology showed an improved R² value compared to the existing CycleGAN-applied image and histogram matching image. For the experiment by patch size used for training, an R² value of about 0.98 was generated for a patch of 1024*1024 pixels. Furthermore, As a result of comparing the pixel distribution for each image band, the simulation result trained with a large patch size showed a more similar histogram distribution to the original image. Therefore, by using ConvNeXt CycleGAN, which is more advanced than the image applied with the existing CycleGAN method and the histogram-matching image, it is possible to derive simulation results similar to the original image and perform a successful simulation.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.3
• /
• pp.11-20
• /
• 2024

Among the generative models in Artificial Intelligence (AI), especially Generative Adversarial Network (GAN) has been successful in various applications such as image processing, density estimation, and style transfer. While the GAN models including Conditional GAN (CGAN), CycleGAN, BigGAN, have been extended and improved, researchers face challenges in real-world applications in specific domains such as disaster simulation, healthcare, and urban planning due to data scarcity and unstable learning causing Image distortion. This paper proposes a new progressive learning methodology called Progressive Step Training (PST) based on the Auxiliary Classifier GAN (ACGAN) that discriminates class labels, leveraging the progressive learning approach of the Progressive Growing of GAN (PGGAN). The PST model achieves 70.82% faster stabilization, 51.3% lower standard deviation, stable convergence of loss values in the later high resolution stages, and a 94.6% faster loss reduction compared to conventional methods.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.40 no.4
• /
• pp.359-364
• /
• 2022

The change detection in remote sensing results in errors due to the camera's optical factors, seasonal factors, and land cover characteristics. The inclination of the building in the image was simulated according to the camera angle using the Cycle Generative Adversarial Network method, and the simulated image was used to contribute to the improvement of change detection accuracy. Based on CycleGAN, the inclination of the building was similarly simulated to the building in the other image based on the image of one of the two periods, and the error of the original image and the inclination of the building was compared and analyzed. The experimental data were taken at different times at different angles, and Kompsat-3A high-resolution satellite images including urban areas with dense buildings were used. As a result of the experiment, the number of incorrect detection pixels per building in the two images for the building area in the image was shown to be reduced by approximately 7 times from 12,632 in the original image and 1,730 in the CycleGAN-based simulation image. Therefore, it was confirmed that the proposed method can reduce detection errors due to the inclination of the building.

• Journal of Ocean Engineering and Technology
• /
• v.34 no.6
• /
• pp.442-453
• /
• 2020

Underwater color images suffer from low visibility and color cast effects caused by light attenuation by water and floating particles. This study applied single image enhancement techniques to enhance the quality of underwater images and compared their performance with real underwater images taken in Korean waters. Dark channel prior (DCP), gradient transform, image fusion, and generative adversarial networks (GAN), such as cycleGAN and underwater GAN (UGAN), were considered for single image enhancement. Their performance was evaluated in terms of underwater image quality measure, underwater color image quality evaluation, gray-world assumption, and blur metric. The DCP saturated the underwater images to a specific greenish or bluish color tone and reduced the brightness of the background signal. The gradient transform method with two transmission maps were sensitive to the light source and highlighted the region exposed to light. Although image fusion enabled reasonable color correction, the object details were lost due to the last fusion step. CycleGAN corrected overall color tone relatively well but generated artifacts in the background. UGAN showed good visual quality and obtained the highest scores against all figures of merit (FOMs) by compensating for the colors and visibility compared to the other single enhancement methods.

• Korean Journal of Radiology
• /
• v.22 no.6
• /
• pp.983-993
• /
• 2021

Objective: To investigate the image quality of ultralow-dose CT (ULDCT) of the chest reconstructed using a cycle-consistent generative adversarial network (CycleGAN)-based deep learning method in the evaluation of pulmonary tuberculosis. Materials and Methods: Between June 2019 and November 2019, 103 patients (mean age, 40.8 ± 13.6 years; 61 men and 42 women) with pulmonary tuberculosis were prospectively enrolled to undergo standard-dose CT (120 kVp with automated exposure control), followed immediately by ULDCT (80 kVp and 10 mAs). The images of the two successive scans were used to train the CycleGAN framework for image-to-image translation. The denoising efficacy of the CycleGAN algorithm was compared with that of hybrid and model-based iterative reconstruction. Repeated-measures analysis of variance and Wilcoxon signed-rank test were performed to compare the objective measurements and the subjective image quality scores, respectively. Results: With the optimized CycleGAN denoising model, using the ULDCT images as input, the peak signal-to-noise ratio and structural similarity index improved by 2.0 dB and 0.21, respectively. The CycleGAN-generated denoised ULDCT images typically provided satisfactory image quality for optimal visibility of anatomic structures and pathological findings, with a lower level of image noise (mean ± standard deviation [SD], 19.5 ± 3.0 Hounsfield unit [HU]) than that of the hybrid (66.3 ± 10.5 HU, p < 0.001) and a similar noise level to model-based iterative reconstruction (19.6 ± 2.6 HU, p > 0.908). The CycleGAN-generated images showed the highest contrast-to-noise ratios for the pulmonary lesions, followed by the model-based and hybrid iterative reconstruction. The mean effective radiation dose of ULDCT was 0.12 mSv with a mean 93.9% reduction compared to standard-dose CT. Conclusion: The optimized CycleGAN technique may allow the synthesis of diagnostically acceptable images from ULDCT of the chest for the evaluation of pulmonary tuberculosis.