Neural Style Transfer for image within images and conditional GANs for destylization

Garg, Mallika; Ubhi, Jagpal Singh; Aggarwal, Ashwani Kumar

doi:10.1016/j.jvcir.2022.103483

Cited by 37 publications

(7 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the early stage of training, the discriminator can often accurately discriminate the false sample data output by the generator, and the generator will quickly fail, so this paper studies the performance control of the discriminator. In order to better control the output performance of the discriminator, it is necessary to satisfy the Lipsey continuity, which is a smoother condition than the usual continuity [11,12]. Specifically, it is desirable to limit the speed at which the function of the discriminator changes, that is, it is necessary to limit the modulus of the function gradient of the discriminator D to fluctuate within a controllable range, and also to make the function smoother.…”

Section: Image Style Migration Methods Based On Cyclegan 321 Normaliz...mentioning

confidence: 99%

See 1 more Smart Citation

Research on Style Transfer of Artistic Images Based on Artificial Intelligence

Wang

2024

Frontiers in Artificial Intelligence and Applications

View full text Add to dashboard Cite

In order to solve some inherent defects of generative confrontation network model in artistic image processing, such as unstable training, gradient disappearance and pattern collapse, the research on artistic image style transfer based on artificial intelligence generative confrontation network is put forward (overall method). On the basis of generative countermeasure network, the network model structure is optimized, and the spectral normalization processing is introduced and the residual structure is improved. Normalization constrains the parameter learning range of the network model, accelerates the learning speed of the model, and has a certain regularization effect; In the generator network, a new residual structure is used to optimize the signal propagation and reduce the training error. Through the treatment of identity mapping, ResNet network alleviates the problems of gradient disappearance and network degradation caused by deep network layers. The experimental results show that 10 subjects, aged between 22 and 30, including 3 females and 7 males, were invited to participate in the test. They had never been exposed to 30 pictures in the two groups, and the corresponding model of each group was unknown. From the results, the stylized pictures using the model proposed in this paper are more popular among young people, with an evaluation rate of 58% and better visual experience. Conclusion: Experiments show that the improved algorithm is better than the original algorithm in image style conversion.

show abstract

Section: Image Style Migration Methods Based On Cyclegan 321 Normaliz...mentioning

confidence: 99%

“…Express. In practical application, it is often used = = = 1Substituting (3-17) into (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16) gives the following formula (20):…”

Section: Experimental Evaluation Criteria and Methodsmentioning

confidence: 99%

Research on Style Transfer of Artistic Images Based on Artificial Intelligence

Wang

2024

Frontiers in Artificial Intelligence and Applications

View full text Add to dashboard Cite

show abstract

“…The training labels of the ROI network were annotated by an otolaryngologist. Inspired by the main ideas of Neural Style Transfer proposed by the researchers Garg [ 11 , 46 ], by combining with the specificity and scarcity of our middle ear MEC data, we reused MEC data, and adding inverted left ear MEC case data to the training of the right ear pathology classifier.…”

Section: Data Description and Preprocessingmentioning

confidence: 99%

Structure-constrained deep feature fusion for chronic otitis media and cholesteatoma identification

Cao

Song

et al. 2023

Multimed Tools Appl

View full text Add to dashboard Cite

Chronic suppurative otitis media (CSOM) and middle ear cholesteatoma (MEC) were two most common chronic middle ear disease(MED) clinically. Accurate differential diagnosis between these two diseases is of high clinical importance given the difference in etiologies, lesion manifestations and treatments. The high-resolution computed tomography (CT) scanning of the temporal bone presents a better view of auditory structures, which is currently regarded as the first-line diagnostic imaging modality in the case of MED. In this paper, we first used a region-of-interest (ROI) network to find the area of the middle ear in the entire temporal bone CT image and segment it to a size of 100*100 pixels. Then, we used a structure-constrained deep feature fusion algorithm to convert different characteristic features of the middle ear in three groups as suppurative otitis media (CSOM), middle ear cholesteatoma (MEC) and normal patches. To fuse structure information, we introduced a graph isomorphism network that implements a feature vector from neighbourhoods and the coordinate distance between vertices. Finally, we construct a classifier named the “otitis media, cholesteatoma and normal identification classifier” (OMCNIC). The experimental results achieved by the graph isomorphism network revealed a 96.36% accuracy in all CSOM and MEC classifications. The experimental results indicate that our structure-constrained deep feature fusion algorithm can quickly and effectively classify CSOM and MEC. It will help otologist in the selection of the most appropriate treatment, and the complications can also be reduced.

show abstract

“…The convolutional neural networks (CNN) have shown great success in artistic Neural Style Transfer (NST) between the various classes of content and style images retrospectively [22]. NST is proliferated in diverse and interesting applications such as image super-resolution [18], geometric warping [42], video [52], CAPTCHA [13], image steganography [44], restorable arbitrary NST [43], etc. Recently, a framed-based arbitrary video style transfer method is proposed by aligning cross-domain features with input videos leveraging multi-channel correlation [35].…”

Section: Introductionmentioning

confidence: 99%