Search

Almost all advanced face swapping approaches use reconstruction as the proxy task, i.e., supervision only exists when the target and source belong to the same person. Otherwise, lacking pixel-level supervision, these methods struggle for source identity preservation. This paper proposes to construct reliable supervision, dubbed cycle triplets, which serves as the image-level guidance when the source identity differs from the target one during training. Specifically, we use face reenactment and blending techniques to synthesize the swapped face from real images in advance, where the synthetic face preserves source identity and target attributes. However, there may be some artifacts in such a synthetic face. To avoid the potential artifacts and drive the distribution of the network output close to the natural one, we reversely take synthetic images as input while the real face as reliable supervision during the training stage of face swapping. Besides, we empirically find that the existing methods tend to lose lower-face details like face shape and mouth from the source. This paper additionally designs a FixerNet, providing discriminative embeddings of lower faces as an enhancement. Our face swapping framework, named ReliableSwap, can boost the performance of any existing face swapping network with negligible overhead. Extensive experiments demonstrate the efficacy of our ReliableSwap, especially in identity preservation. The project page is https://reliable-swap.github.io/., Comment: Project page: https://reliable-swap.github.io/ ; Github repository: https://github.com/ygtxr1997/ReliableSwap ; Demo (HuggingFace): https://huggingface.co/spaces/ygtxr1997/ReliableSwap_Demo

Matching-based methods, especially those based on space-time memory, are significantly ahead of other solutions in semi-supervised video object segmentation (VOS). However, continuously growing and redundant template features lead to an inefficient inference. To alleviate this, we propose a novel Sequential Weighted Expectation-Maximization (SWEM) network to greatly reduce the redundancy of memory features. Different from the previous methods which only detect feature redundancy between frames, SWEM merges both intra-frame and inter-frame similar features by leveraging the sequential weighted EM algorithm. Further, adaptive weights for frame features endow SWEM with the flexibility to represent hard samples, improving the discrimination of templates. Besides, the proposed method maintains a fixed number of template features in memory, which ensures the stable inference complexity of the VOS system. Extensive experiments on commonly used DAVIS and YouTube-VOS datasets verify the high efficiency (36 FPS) and high performance (84.3\% $\mathcal{J}\&\mathcal{F}$ on DAVIS 2017 validation dataset) of SWEM. Code is available at: https://github.com/lmm077/SWEM., 15 pages with Supplementary Material

We present a novel paradigm for high-fidelity face swapping that faithfully preserves the desired subtle geometry and texture details. We rethink face swapping from the perspective of fine-grained face editing, \textit{i.e., ``editing for swapping'' (E4S)}, and propose a framework that is based on the explicit disentanglement of the shape and texture of facial components. Following the E4S principle, our framework enables both global and local swapping of facial features, as well as controlling the amount of partial swapping specified by the user. Furthermore, the E4S paradigm is inherently capable of handling facial occlusions by means of facial masks. At the core of our system lies a novel Regional GAN Inversion (RGI) method, which allows the explicit disentanglement of shape and texture. It also allows face swapping to be performed in the latent space of StyleGAN. Specifically, we design a multi-scale mask-guided encoder to project the texture of each facial component into regional style codes. We also design a mask-guided injection module to manipulate the feature maps with the style codes. Based on the disentanglement, face swapping is reformulated as a simplified problem of style and mask swapping. Extensive experiments and comparisons with current state-of-the-art methods demonstrate the superiority of our approach in preserving texture and shape details, as well as working with high resolution images. The project page is http://e4s2022.github.io, Comment: Accepted to CVPR 2023

In this paper, considering the various terminal states constraints, and coupling of different states in the ascent phase of the hypersonic aircraft, a multi-constrained guidance method based on predictor–corrector policy is proposed. Firstly, the dynamic equations of ascent phase are given. And then, the dynamic amplification factors for the various flight states are calculated. Analyzing the dynamic equations of ascent phase, it can be found that only the attack angle is used to control various states and the guidance of ascent phase is an under-driven problem. The path angles are small values at the terminal flight phase, and it has little couple influence on the altitude. In order to solve the under-driven problem, the whole ascent phase has two parts. At the initial phase, eliminating the terminal height error is chosen as the guidance object. At the terminal phase, eliminating the path angle error at the end point is chosen as the guidance object. What’s more, the influence of choosing different segmentation altitude on the terminal states errors is analyzed. Moreover, the Monte Carlo simulation demonstrates the effectiveness of the proposed multi-constrained adaptive predictor–corrector guidance method.

Dongxiang common wild rice (Oryza rufipogon Griff.) is believed to have the northernmost distribution of any wild rice species. Many favorable genes were lost during rice domestication, and Dongxiang common wild rice is a potential source of many genes related to biological and abiotic stress resistance and high grain yield. Despite its importance, japonica rice has not yet been used as a background material for the construction of introgression lines carrying Dongxiang common wild rice chromosome segments. In this study, we used Dongxiang common wild rice as the donor parent and Nipponbare (Oryza sativa L), a reference-sequence japonica cultivar, as the recurrent parent to develop a set of 104 chromosome segment substitution lines (CSSLs) through crossing and backcrossing with marker-assisted selection based on 203 polymorphic molecular markers evenly distributed across 12 chromosomes. The 104 CSSLs covered 87.94% of the genome of Dongxiang common wild rice. Each CSSL contained an average of four introgressed segments, and the average segment length was 3.3 Mb. Quantitative trait locus (QTL) analysis of the panicle-related traits of the CSSLs at two sites (Beijing and Nanchang, Jiangxi Province) revealed 18 QTLs for eight traits. Among these 18 QTLs were two known grain length- and width-related genes and four novel QTLs. In addition, two QTLs were verified, and two novel QTLs were identified, for panicle neck length, a domestication-related trait. The developed CSSLs potentially represent a valuable population for detecting favorable genes in common wild rice and creating germplasm resources for japonica rice breeding.

The ongoing rise in temperatures caused by global climate change is a critical climatic risk factor for rice production, and enhancing rice heat tolerance is an area of particular research interest. A recombinant inbred line (RIL) mapping population was developed from heat sensitive, rice cultivar IAPAR-9 crossed with heat tolerant, Liaoyan241. RIL and parental lines were exposed to high temperature at the heating and flowering stage in experiments in 2014 and 2015. As indicators of heat tolerance, the seed setting rate under natural (NS) and heat stress (HTS) conditions were measured, and the reduction rate of seed set (RRS) was calculated. Quantitative trait loci (QTL) analysis revealed eleven heat tolerance QTLs located on chromosomes 1, 3, 4, 5, and 6. Single QTL contribution rates were 4.75–13.81% and effect values were − 5.98 to 5.00. Four major QTLs (qNS1, qNS4, qNS6, and qRRS1) were stable detected in different environments in both years. Thirteen QTLs with epistatic interactions and nine QTLs with environmental interactions were also detected. Major QTLs were all involved in epistatic and environmental interactions. Three QTLs from the SSR marker interval RM471 to RM177 region of chromosome 4 (qNS4, qHTS4, and qRRS4) were all involved in epistatic and environmental interactions and contributed to phenotypic variation, indicating that this region constituted a major QTL hotspot. The major QTL for heat tolerance identified in this study will aid in breeding tolerant cultivars and facilitating investigation of the molecular underpinnings of heat tolerance in rice.