Review on Generative Adversarial Networks: Focusing on Computer Vision and Its Applications

Abstract

The emergence of deep learning model GAN (Generative Adversarial Networks) is an important turning point in generative modeling. GAN is more powerful in feature and expression learning compared to machine learning-based generative model algorithms. Nowadays, it is also used to generate non-image data, such as voice and natural language. Typical technologies include BERT (Bidirectional Encoder Representations from Transformers), GPT-3 (Generative Pretrained Transformer-3), and MuseNet. GAN differs from the machine learning-based generative model and the objective function. Training is conducted by two networks: generator and discriminator. The generator converts random noise into a true-to-life image, whereas the discriminator distinguishes whether the input image is real or synthetic. As the training continues, the generator learns more sophisticated synthesis techniques, and the discriminator grows into a more accurate differentiator. GAN has problems, such as mode collapse, training instability, and lack of evaluation matrix, and many researchers have tried to solve these problems. For example, solutions such as one-sided label smoothing, instance normalization, and minibatch discrimination have been proposed. The field of application has also expanded. This paper provides an overview of GAN and application solutions for computer vision and artificial intelligence healthcare field researchers. The structure and principle of operation of GAN, the core models of GAN proposed to date, and the theory of GAN were analyzed. Application examples of GAN such as image classification and regression, image synthesis and inpainting, image-to-image translation, super-resolution and point registration were then presented. The discussion tackled GAN’s problems and solutions, and the future research direction was finally proposed.

Additional Information Delivery to Image Content via Improved Unseen–Visible Watermarking

“…Considering Table 4 and several aspects of each algorithm shown in Table 5, the proposal is more suitable to be used in practical applications of additional information delivery, information security tasks [15][16][17][18][19] and other related fields [20], such as is mentioned in Section 1.…”

“…However, when µ R ≥ 253 and JND(µ R ) = 3, an overflow can be generated by the embedding strategy into the pixels of the watermarked region Rw. In this way, to avoid this overflow, the embedding strength is finally obtained by a soft adjustment, as is formulated in (17).…”

“…In this proposal, additional information delivery to the image content refers to the application of the general scenario of the UVW algorithms in daily life; when the endusers exhibit the QR watermark, the decoded information could be a company website, information of singers, dress designer, location or another kind of information that has or does not have relation to the image content. Additionally, it is not limited to extend the applicability to information security tasks such as medical imaging [15,16], deepfake prevention [17], 3D-video protection [18], face recognition authentication [19] and other related fields [20]. However, to obtain better results in all those applications, several drawbacks of the UVW algorithms must be improved: first, the algorithms only require large image regions with low intensity [6][7][8]14] second, embedding strategies increase histogram distortion and visual degradation of the watermarked image, impacting the legibility of the visual content and the watermark imperceptibility.…”

Music Composition Using Quantum Annealing

“…MuseNet also combines several different musical styles, including those of Mozart, Beatles, and Country music [59]. These techniques are being further refined with the use of Generative Adversarial Networks (GAN) and the development of General Purpose Transformers (GPT-2 and GPT-3) [60].…”

A Hybrid Imputation Method for Multi-Pattern Missing Data: A Case Study on Type II Diabetes Diagnosis

“…Li et al [110] proposed a novel hybrid method coupling empirical mode decomposition and a long short-term memory deep learning network to predict missing measured signal data of structural health monitoring (SHM) systems. The generative adversarial network is the next frontier of machine learning [111] which is applied in the machine learning data imputation approach and has the potential to handle missing data accurately and efficiently. Zhang et al [112] proposed a model of end-to-end generative adversarial network with real-data forcing to impute the missing values in a multivariate time series.…”