SDP-GAN: Saliency Detail Preservation Generative Adversarial Networks for High Perceptual Quality Style Transfer

USOD10K: A New Benchmark Dataset for Underwater Salient Object Detection

Underwater salient object detection (USOD) is an emerging research area that has great potential for various underwater visual tasks. However, USOD research is still in its early stage due to the lack of large-scale datasets within which salient objects are well-defined and pixel-wise annotated. To address this issue, this paper introduces a new dataset named USOD10K. It contains 10,255 underwater images, covering 70 categories of salient objects in 12 different underwater scenes. Moreover, the USOD10K provides salient object boundaries and depth maps of all images. The USOD10K is the first large-scale dataset in the USOD community, making a significant leap in diversity, complexity, and scalability. Secondly, a simple but strong baseline termed TC-USOD is proposed for the USOD10K. The TC-USOD adopts a hybrid architecture based on an encoder-decoder design that leverages transformer and convolution as the basic computational building block of the encoder and decoder, respectively. Thirdly, we make a comprehensive summarization of 35 state-of-the-art SOD/USOD methods and benchmark them on the existing USOD dataset and the USOD10K. The results show that our TC-USOD achieves superior performance on all datasets tested. Finally, several other use cases of the USOD10K are discussed, and future directions of USOD research are pointed out. This work will promote the development of the USOD research and facilitate further research on underwater visual tasks and visually-guided underwater robots. To pave the road in the USOD research field, the dataset, code, and benchmark results are publicly available: https://github.com/Underwater-Robotic-Lab/USOD10K.

Application of style transfer algorithm in the integration of traditional garden and modern design elements

With the development of society, modern design elements are increasingly integrated into traditional garden design, forming a novel style fusion that improves both aesthetics and the sustainability of the social-ecological system. This study explores the application of style transfer algorithms to seamlessly integrate the aesthetics of traditional landscape paintings with virtual scenes of classical private gardens. The effectiveness of the method is verified through a series of experiments using virtual scenes of the Humble Administrator's Garden and various landscape paintings representing different artistic styles. The experimental results demonstrate that the style transfer technique can accurately replicate the aesthetic features of traditional paintings and integrate them into the virtual garden environment. This approach highlights the potential of combining cultural heritage with advanced technological methods, indicating that the technology has great potential to innovate garden design by promoting the synergy between cultural heritage and technological innovation. By promoting the integration of traditional aesthetics and modern design principles, we contribute to the sustainability and richness of the social-ecological system and provide a framework for future digital preservation and restoration applications of urban cultural heritage. The code for implementing TRD-Net is available at https://github.com/huangbei029/Hybrid-Garden-StyleNet-dd/tree/main.

Saliency Guided Deep Neural Network for Color Transfer With Light Optimization

Color transfer aims to change the color information of the target image according to the reference one. Many studies propose color transfer methods by analysis of color distribution and semantic relevance, which do not take the perceptual characteristics for visual quality into consideration. In this study, we propose a novel color transfer method based on the saliency information with brightness optimization. First, a saliency detection module is designed to separate the foreground regions from the background regions for images. Then a dual-branch module is introduced to implement color transfer for images. Finally, a brightness optimization operation is designed during the fusion of foreground and background regions for color transfer. Experimental results show that the proposed method can implement the color transfer for images while keeping the color consistency well. Compared with other existing studies, the proposed method can obtain significant performance improvement. The source code and pre-trained models are available at https://github.com/PlanktonQAQ/SCTNet.

GAN-FDSR: GAN-Based Fault Detection and System Reconfiguration Method

Fault detection and exclusion are essential to ensure the integrity and reliability of the tightly coupled global navigation satellite system (GNSS)/inertial navigation system (INS) integrated navigation system. A fault detection and system reconfiguration scheme based on generative adversarial networks (GAN-FDSR) for tightly coupled systems is proposed in this paper. The chaotic characteristics of pseudo-range data are analyzed, and the raw data are reconstructed in phase space to improve the learning ability of the models for non-linearity. The trained model is used to calculate generation and discrimination scores to construct fault detection functions and detection thresholds while retaining the generated data for subsequent system reconfiguration. The influence of satellites on positioning accuracy of the system under different environments is discussed, and the system reconfiguration scheme is dynamically selected by calculating the relative differential precision of positioning (RDPOP) of the faulty satellites. Simulation experiments are conducted using the field test data to assess fault detection performance and positioning accuracy. The results show that the proposed method greatly improves the detection sensitivity of the system for small-amplitude faults and gradual faults, and effectively reduces the positioning error during faults.

CartoonLossGAN: Learning Surface and Coloring of Images for Cartoonization

Cartoonization as a special type of artistic style transfer is a difficult image processing task. The current existing artistic style transfer methods cannot generate satisfactory cartoon-style images due to that artistic style images often have delicate strokes and rich hierarchical color changes while cartoon-style images have smooth surfaces without obvious color changes, and sharp edges. To this end, we propose a cartoon loss based generative adversarial network (CartoonLossGAN) for cartoonization. Particularly, we first reuse the encoder part of the discriminator to build a compact generative adversarial network (GAN) based cartoonization architecture. Then we propose a novel cartoon loss function for the architecture. It can imitate the process of sketching to learn the smooth surface of the cartoon image, and imitate the coloring process to learn the coloring of the cartoon image. Furthermore, we also propose an initialization strategy, which is used in the scenario of reusing the discriminator to make our model training easier and more stable. Extensive experimental results demonstrate that our proposed CartoonLossGAN can generate fantastic cartoon-style images, and outperforms four representative methods.