Real-World Light Field Image Super-Resolution Via Degradation Modulation.

Recent years have witnessed the great advances of deep neural networks (DNNs) in light field (LF) image super-resolution (SR). However, existing DNN-based LF image SR methods are developed on a single fixed degradation (e.g., bicubic downsampling), and thus cannot be applied to super-resolve real LF images with diverse degradation. In this article, we propose a simple yet effective method for real-world LF image SR. In our method, a practical LF degradation model is developed to formulate the degradation process of real LF images. Then, a convolutional neural network is designed to incorporate the degradation prior into the SR process. By training on LF images using our formulated degradation, our network can learn to modulate different degradation while incorporating both spatial and angular information in LF images. Extensive experiments on both synthetically degraded and real-world LF images demonstrate the effectiveness of our method. Compared with existing state-of-the-art single and LF image SR methods, our method achieves superior SR performance under a wide range of degradation, and generalizes better to real LF images. Codes and models are available at https://yingqianwang.github.io/LF-DMnet/.

Latent profiles of bullying perpetration and victimization: Gender differences and family variables.

BACKGROUND: School bullying is a prevalent issue that threatens the psychological and social well-being of adolescents. However, little research has investigated how gender and family variables were related to bullying-involvement patterns among adolescents with siblings. OBJECTIVE: This study explored gender differences in the profiles of bullying involvement and the relationship between sibling, parental variables, and these profiles among Chinese adolescents. PARTICIPANTS AND SETTING: Participants (N = 1,060; 46.0 % boys; Mage = 15.53) were recruited from junior and senior high schools in Jiangxi and Guizhou Provinces, China. METHODS: Bullying involvement, sibling warmth and conflict, and parental psychological maltreatment and neglect were assessed by self-report questionnaire. Latent profile analysis was used to identify subgroups with distinct bullying involvement patterns, then multiple logistic regressions were performed to investigate the associations between family variables and bullying-involvement subgroups. RESULTS: We found gender differences in both the latent profiles of bullying involvement and the associations between profiles and family variables. Only boys were identified severe bully-victims (3.39 %), while only girls were categorized as relational bully-victims (20.18 %). Boys and girls were similarly represented among uninvolved students (70.76 % vs. 66.85 %), moderate bully-victims (15.25 % vs. 6.49 %), and victims (10.59 % vs. 6.49 %). Students with more sibling warmth manifested less likelihood of engaging in bullying-related profiles, with more parental psychological maltreatment, and more parental neglect manifested more likelihood of engaging in bullying-related profiles only among girls. While students with more sibling conflict were related to more bullying-related profiles among boys than girls. CONCLUSIONS: The findings emphasize the importance of developing gender-specific bullying intervention strategies that also consider relevant family factors.

[Retracted] Long non­coding RNA SNHG6 promotes tumorigenesis in melanoma cells via the microRNA­101­3p/RAP2B axis.

[This retracts the article DOI: 10.3892/ol.2020.12186.].

Artesunate Induces Ferroptosis in Hepatic Stellate Cells and Alleviates Liver Fibrosis via the ROCK1/ATF3 Axis.

Background and Aims: Development of fibrosis in chronic liver disease requires activation of hepatic stellate cells (HSCs) and leads to a poor outcome. Artesunate (Art) is an ester derivative of artemisinin that can induce ferroptosis in HSCs, and activated transcriptional factor 3 (ATF3) is an ATF/CREB transcription factor that is induced in response to stress. In this study, we examined the role of the Rho-associated protein kinase 1 (ROCK1)/ATF3 axis in Art-induced ferroptosis in HSCs. Methods: HSC activation and ferroptosis were studied in vitro by western blotting, polymerase chain reaction, immunofluorescence, and other assays. ATF3 electrophoretic mobility and ROCK1 protein stability were assayed by western blotting. Immunoprecipitation was used to detect the interaction of ROCK1 and ATF3, as well as ATF3 phosphorylation. A ubiquitination assay was used to verify ROCK1 degradation. Atf3-interfering and Rock1-overexpressing mice were constructed to validate the anti-hepatic fibrosis activity of Art in vivo. Results: Art induced ferroptosis in HSCs following glutathione-dependent antioxidant system inactivation resulting from nuclear accumulation of unphosphorylated ATF3 mediated by ROCK1-ubiquitination in vitro. Art also decreased carbon tetrachloride-induced liver fibrosis in mice, which was reversed by interfering with Atf3 or overexpressing Rock1. Conclusions: The ROCK1/ATF3 axis was involved in liver fibrosis and regulation of ferroptosis, which provides an experimental basis for further study of Art for the treatment of liver fibrosis.

Autophagy activation is required for N6-methyladenosine modification to regulate ferroptosis in hepatocellular carcinoma.

BACKGROUND & AIMS: Although ferroptosis holds promise as a new strategy for treating hepatocellular carcinoma (HCC), there are several obstacles that need to be overcome. One major challenge is the lack of understanding about the mechanisms underlying ferroptosis. Additionally, while the m6A modification has been shown to regulate various forms of cell death, its role in regulating ferroptosis in HCC has been largely overlooked. Bridging this knowledge gap, our study aimed to elucidate the regulatory influence of m6A modification on HCC ferroptosis. MATERIALS: Dot blot and EpiQuik m6A RNA Methylation Quantitative kit detected changes in overall m6A modification level during ferroptosis in HCC. MeRIP-qPCR and RIP-qPCR identified that the m6A modification of ATG5 mRNA was significant changed. BALB/c nude mice were used to construct xenograft tumor models to verify the phenotypes upon YTHDC2 silencing. In addition, patient-derived organoid models were used to demonstrate that induction of ferroptosis was an effective strategy against HCC. RESULTS: Our study has shown that inducing ferroptosis is a promising strategy for combatting HCC. Specifically, we have found a significant correlation between ferroptosis and high levels of m6A modification in HCC. Notably, we discovered that the elevation of ATG5 mRNA m6A modification mediated by WTAP is dependent on the reading protein YTHDC2. Importantly, inhibition of either WTAP or YTHDC2 effectively prevented ferroptosis and suppressed HCC development in both in vitro and in vivo models. CONCLUSION: Our study revealed that WTAP upregulates ATG5 expression post-transcriptionally in an m6A-YTHDC2-dependent manner, thereby promoting the translation of ATG5 mRNA during ferroptosis in HCC. These findings have significant implications for the development of innovative and effective therapeutic approaches for HCC treatment.

Direction-Coded Temporal U-Shape Module for Multiframe Infrared Small Target Detection.

Infrared small target (IRST) detection aims at separating targets from cluttered background. Although many deep learning-based single-frame IRST (SIRST) detection methods have achieved promising detection performance, they cannot deal with extremely dim targets while suppressing the clutters since the targets are spatially indistinctive. Multiframe IRST (MIRST) detection can well handle this problem by fusing the temporal information of moving targets. However, the extraction of motion information is challenging since general convolution is insensitive to motion direction. In this article, we propose a simple yet effective direction-coded temporal U-shape module (DTUM) for MIRST detection. Specifically, we build a motion-to-data mapping to distinguish the motion of targets and clutters by indexing different directions. Based on the motion-to-data mapping, we further design a direction-coded convolution block (DCCB) to encode the motion direction into features and extract the motion information of targets. Our DTUM can be equipped with most single-frame networks to achieve MIRST detection. Moreover, in view of the lack of MIRST datasets, including dim targets, we build a multiframe infrared small and dim target dataset (namely, NUDT-MIRSDT) and propose several evaluation metrics. The experimental results on the NUDT-MIRSDT dataset demonstrate the effectiveness of our method. Our method achieves the state-of-the-art performance in detecting infrared small and dim targets and suppressing false alarms. Our codes will be available at https://github.com/TinaLRJ/Multi-frame-infrared-small-target-detection-DTUM.

COVID-19-associated liver injury: Adding fuel to the flame.

COVID-19 is mainly characterized by respiratory disorders and progresses to multiple organ involvement in severe cases. With expansion of COVID-19 and SARS-CoV-2 research, correlative liver injury has been revealed. It is speculated that COVID-19 patients exhibited abnormal liver function, as previously observed in the SARS and MERS pandemics. Furthermore, patients with underlying diseases such as chronic liver disease are more susceptible to SARS-CoV-2 and indicate a poor prognosis accompanied by respiratory symptoms, systemic inflammation, or metabolic diseases. Therefore, COVID-19 has the potential to impair liver function, while individuals with preexisting liver disease suffer from much worse infected conditions. COVID-19 related liver injury may be owing to direct cytopathic effect, immune dysfunction, gut-liver axis interaction, and inappropriate medication use. However, discussions on these issues are infancy. Expanding research have revealed that angiotensin converting enzyme 2 (ACE2) expression mediated the combination of virus and target cells, iron metabolism participated in the virus life cycle and the fate of target cells, and amino acid metabolism regulated immune response in the host cells, which are all closely related to liver health. Further exploration holds great significance in elucidating the pathogenesis, facilitating drug development, and advancing clinical treatment of COVID-19-related liver injury. This article provides a review of the clinical and laboratory hepatic characteristics in COVID-19 patients, describes the etiology and impact of liver injury, and discusses potential pathophysiological mechanisms.

Surface ligand-assisted synthesis and biomedical applications of metal-organic framework nanocomposites.

Metal-organic framework (MOF) nanocomposites have recently gained intensive attention for biosensing and disease therapy applications owing to their outstanding physiochemical properties. However, the direct growth of MOF nanocomposites is usually hindered by the mismatched lattice in the interface between the MOF and other nanocomponents. Surface ligands, molecules with surfactant-like properties, are demonstrated to exhibit the robust capability to modify the interfacial properties of nanomaterials and can be utilized as a powerful strategy for the synthesis of MOF nanocomposites. Besides this, surface ligands also exhibit significant functions in the morphological control and functionalization of MOF nanocomposites, thus greatly enhancing their performance in biomedical applications. In this review, the surface ligand-assisted synthesis and biomedical applications of MOF nanocomposites are comprehensively reviewed. Firstly, the synthesis of MOF nanocomposites is discussed according to the diverse roles of surface ligands. Then, MOF nanocomposites with different properties are listed with their applications in biosensing and disease therapy. Finally, current challenges and further directions of MOF nanocomposites are presented to motivate the development of MOF nanocomposites with elaborate structures, enriched functions, and excellent application prospects.

Perceptions of Parental Privacy Invasion and Information Management among Chinese Adolescents: Comparing Between- and Within-Family Associations.

The traditional Chinese conceptualization of family privacy is interdependent and hierarchically structured, but mounting evidence suggests that contemporary Chinese youth hold strong desires for individual privacy and respond defensively to perceived parental privacy invasions. The current research examined within-person associations among adolescents' perceptions of parental privacy invasion, secrecy, and disclosure to parents in the Chinese context. This study collected data from 289 Chinese youth (MageT1 = 13.57, SD = 0.63, 50.30% male) at six-month intervals over one year. Random intercept cross-lagged panel modeling (RI-CLPM) showed that stronger perceptions of parental invasion predicted later within-person decreases in adolescents' disclosure and increases in secrecy. Disclosure and secrecy did not predict later perceptions of parental invasion at the within-person level. The findings suggest that Chinese youth manage privacy reactively and defensively when feelings of invasion occur, by decreasing disclosure and increasing secrecy. Stereotypes portraying Chinese youth as highly deferential to parents' demands for informational access might not be representative of adolescents in contemporary society.

Longitudinal links between expressive flexibility and friendship quality in adolescence: The moderating effect of social anxiety.

INTRODUCTION: Expressive flexibility, or the ability to both up- and down-regulate emotional expressions in social interactions, is thought as an indicator and a consequence of healthy interpersonal relationships. The present longitudinal study examined bidirectional associations between expressive flexibility and friendship quality in early adolescence. Since prior research found inconsistent results regarding the adaptiveness of expressive flexibility, which indicated the necessity to consider individual variability in the process, we further tested the potential moderating effect of social anxiety in the links from expressive flexibility to friendship quality. METHODS: Participants from two junior high schools in eastern China (N = 274; 50.4% female; Mage = 13.56) were surveyed at three time points with 6-month intervals. Expressive flexibility, friendship quality, and social anxiety were all assessed via self-reported scales. RESULTS: According to the cross-lagged model results, friendship quality significantly predicted increased expressive flexibility over time. Conversely, the longitudinal association from expressive flexibility to friendship quality was not significant, but the interaction between expressive flexibility and social anxiety significantly predicted later friendship quality. Further analyses via the Johnson-Neyman technique revealed that expressive flexibility only positively predicted friendship quality for adolescents with lower levels of social anxiety. CONCLUSION: Our results suggest that expressive flexibility is not always socially adaptive, so practical interventions that aim to improve youths' social adjustment via expressive flexibility training might need to consider the role of individual characteristics.

Comparisons of violence exposure within and across contexts in predicting depression among Chinese children and adolescents.

BACKGROUND: Existing research has identified violence exposure as a risk factor for depression among children and adolescents. To date, however, we know little about whether these associations are influenced by the forms (witnessing and victimization) of violence exposure across different contexts (family, school, and community). OBJECTIVE: The present study thus aimed to compare the effects of two forms of violence exposure across and within three contexts. Sex and age differences were also tested in the above associations. PARTICIPANTS AND SETTING: Participants (N = 906, 52.4 % boys, T1: Mage = 11.03) were children and adolescents recruited from elementary and junior schools in Zhejiang Province, China. METHODS: All participants were asked to complete questionnaires about their experiences of various types of violence exposure and depression at two time points with a 12-month interval. SPSS and Mplus were used to analyze the data. RESULTS: Considering the comparison within contexts, witnessing family violence was a more influential predictor than family violence victimization while school violence was the opposite. As for the comparison across contexts, no significant differences were found in the effects of different violence victimization, while the effect of witnessing family violence and community violence were more influential than witnessing school violence. No significant sex differences in the above associations were found. However, community violence victimization was more negatively related to depression among adolescents compared with children. CONCLUSIONS: The associations between violence exposure and depression are influenced by the forms and contexts of violence as well as age groups.

Oroxylin A regulates cGAS DNA hypermethylation induced by methionine metabolism to promote HSC senescence.

Relevant studies have recognized the important role of hepatic stellate cell (HSC) senescence in anti-liver fibrosis. Cellular senescence is believed to be regulated by the cGAS-STING signaling pathway. However, underlying exact mechanisms of cGAS-STING pathway in hepatic stellate cell senescence are still unclear. Here, we found that Oroxylin A could promote senescence in HSC by activating the cGAS-STING pathway. Moreover, activation of the cGAS-STING pathway was dependent on DNMT3A downregulation, which suppressed cGAS gene DNA methylation. Interestingly, the attenuation of DNMT activity relied on the reduction of methyl donor SAM level. Noteworthy, the downregulation of SAM levels implied the imbalance of methionine cycle metabolism, and MAT2A was considered to be an important regulatory enzyme in metabolic processes. In vivo experiments also indicated that Oroxylin A induced senescence of HSCs in mice with liver fibrosis, and DNMT3A overexpression partly offset this effect. In conclusion, we discovered that Oroxylin A prevented the methylation of the cGAS gene by preventing the production of methionine metabolites, which promoted the senescence of HSCs. This finding offers a fresh hypothesis for further research into the anti-liver fibrosis mechanism of natural medicines.

Exploring Fine-Grained Sparsity in Convolutional Neural Networks for Efficient Inference.

Neural networks contain considerable redundant computation, which drags down the inference efficiency and hinders the deployment on resource-limited devices. In this paper, we study the sparsity in convolutional neural networks and propose a generic sparse mask mechanism to improve the inference efficiency of networks. Specifically, sparse masks are learned in both data and channel dimensions to dynamically localize and skip redundant computation at a fine-grained level. Based on our sparse mask mechanism, we develop SMPointSeg, SMSR, and SMStereo for point cloud semantic segmentation, single image super-resolution, and stereo matching tasks, respectively. It is demonstrated that our sparse masks are well compatible to different model components and network architectures to accurately localize redundant computation, with computational cost being significantly reduced for practical speedup. Extensive experiments show that our SMPointSeg, SMSR, and SMStereo achieve state-of-the-art performance on benchmark datasets in terms of both accuracy and efficiency.

Disentangling Light Fields for Super-Resolution and Disparity Estimation.

Light field (LF) cameras record both intensity and directions of light rays, and encode 3D scenes into 4D LF images. Recently, many convolutional neural networks (CNNs) have been proposed for various LF image processing tasks. However, it is challenging for CNNs to effectively process LF images since the spatial and angular information are highly inter-twined with varying disparities. In this paper, we propose a generic mechanism to disentangle these coupled information for LF image processing. Specifically, we first design a class of domain-specific convolutions to disentangle LFs from different dimensions, and then leverage these disentangled features by designing task-specific modules. Our disentangling mechanism can well incorporate the LF structure prior and effectively handle 4D LF data. Based on the proposed mechanism, we develop three networks (i.e., DistgSSR, DistgASR and DistgDisp) for spatial super-resolution, angular super-resolution and disparity estimation. Experimental results show that our networks achieve state-of-the-art performance on all these three tasks, which demonstrates the effectiveness, efficiency, and generality of our disentangling mechanism. Project page: https://yingqianwang.github.io/DistgLF/.

Dense Nested Attention Network for Infrared Small Target Detection.

Single-frame infrared small target (SIRST) detection aims at separating small targets from clutter backgrounds. With the advances of deep learning, CNN-based methods have yielded promising results in generic object detection due to their powerful modeling capability. However, existing CNN-based methods cannot be directly applied to infrared small targets since pooling layers in their networks could lead to the loss of targets in deep layers. To handle this problem, we propose a dense nested attention network (DNA-Net) in this paper. Specifically, we design a dense nested interactive module (DNIM) to achieve progressive interaction among high-level and low-level features. With the repetitive interaction in DNIM, the information of infrared small targets in deep layers can be maintained. Based on DNIM, we further propose a cascaded channel and spatial attention module (CSAM) to adaptively enhance multi-level features. With our DNA-Net, contextual information of small targets can be well incorporated and fully exploited by repetitive fusion and enhancement. Moreover, we develop an infrared small target dataset (namely, NUDT-SIRST) and propose a set of evaluation metrics to conduct comprehensive performance evaluation. Experiments on both public and our self-developed datasets demonstrate the effectiveness of our method. Compared to other state-of-the-art methods, our method achieves better performance in terms of probability of detection ( Pd ), false-alarm rate ( Fa ), and intersection of union ( IoU ).

Proto-oncogene FAM83A contributes to casein kinase 1-mediated mitochondrial maintenance and white adipocyte differentiation.

Family with sequence similarity 83 A (FAM83A) is a newly discovered proto-oncogene that has been shown to play key roles in various cancers. However, the function of FAM83A in other physiological processes is not well known. Here, we report a novel function of FAM83A in adipocyte differentiation. We used an adipocyte-targeting fusion oligopeptide (FITC-ATS-9R) to deliver a FAM83A-sgRNA/Cas9 plasmid to knockdown Fam83a (ATS/sg-FAM83A) in white adipose tissue in mice, which resulted in reduced white adipose tissue mass, smaller adipocytes, and mitochondrial damage that was aggravated by a high-fat diet. In cultured 3T3-L1 adipocytes, we found loss or knockdown of Fam83a significantly repressed lipid droplet formation and downregulated the expression of lipogenic genes and proteins. Furthermore, inhibition of Fam83a decreased mitochondrial ATP production through blockage of the electron transport chain, associated with enhanced apoptosis. Mechanistically, we demonstrate FAM83A interacts with casein kinase 1 (CK1) and promotes the permeability of the mitochondrial outer membrane. Furthermore, loss of Fam83a in adipocytes hampered the formation of the TOM40 complex and impeded CK1-driven lipogenesis. Taken together, these results establish FAM83A as a critical regulator of mitochondria maintenance during adipogenesis.

Exposure to domestic violence and depressive symptoms in Chinese adolescents: Sleep problems as a mediator.

INTRODUCTION: Exposure to domestic violence has been shown to be an important risk factor of depression in western studies, but has received less attention in Chinese context. Additionally, the underlying mechanism of its link with depression has not been fully studied. With a longitudinal design, we examined the mediating role of sleep problems between exposure to domestic violence and depressive symptoms in Chinese adolescents, and further considered potential age and gender differences in the direct and indirect pathways. METHODS: Participants were recruited from primary and junior high schools in China and were surveyed across two waves with a 6-month interval. There were 1949 participants at Wave 1 and 1283 at Wave 2. Structural equation model was conducted to examine the mediating role of sleep problems in the association between exposure to domestic violence and depressive symptoms. Multigroup analyses were applied to test potential age and gender differences in the process. RESULTS: Participants' exposure to domestic violence predicted increasing sleep problems, which further predicted more depressive symptoms. Age and gender moderated the indirect pathway from exposure to domestic violence to depressive symptoms through sleep problems, such that the mediating effects of sleep problems were significantly stronger in early-age adolescents and girls than middle-age ones and boys. IMPLICATION: Parents should avoid the use of corporal punishment and protect their children from domestic violence. For adolescents who have underwent domestic violence, parents and clinicians might consider to decrease their depressive symptoms by ameliorating sleep problems, especially for the early-age adolescents and girls.

Inflorescence Transcriptome Sequencing and Development of New EST-SSR Markers in Common Buckwheat (Fagopyrum esculentum).

Common buckwheat (Fagopyrum esculentum M.) is known for its adaptability, good nutrition, and medicinal and health care value. However, genetic studies of buckwheat have been hindered by limited genomic resources and genetic markers. In this study, Illumina HiSeq 4000 high-throughput sequencing technology was used to sequence the transcriptome of green-flower common buckwheat (Gr) with coarse pedicels and white-flower Ukrainian daliqiao (UD) with fine pedicels. A total of 118,448 unigenes were obtained, with an average length of 1248 bp and an N50 of 1850 bp. A total of 39,432 differentially expressed genes (DEGs) were identified, and the DEGs of the porphyrins and chlorophyll metabolic pathway had significantly upregulated expression in Gr. Then, a total of 17,579 sequences containing SSR loci were detected, and 20,756 EST-SSR loci were found. The distribution frequency of EST-SSR in the transcriptome was 17.52%, and the average distribution density was 8.21 kb. A total of 224 pairs of primers were randomly selected for synthesis; 35 varieties of common buckwheat and 13 varieties of Tartary buckwheat were verified through these primers. The clustering results well verified the previous conclusion that common buckwheat and Tartary buckwheat had a distant genetic relationship. The EST-SSR markers identified and developed in this study will be helpful to enrich the transcriptome information and marker-assisted selection breeding of buckwheat.

m6A methylation is required for dihydroartemisinin to alleviate liver fibrosis by inducing ferroptosis in hepatic stellate cells.

Activation of hepatic stellate cells (HSCs) is a central event in the development of liver fibrosis, and the elimination of activated HSCs is considered to be an effective anti-fibrotic strategy. Here, we report that dihydroartemisinin (DHA) prevented the activation of HSCs via ferroptosis pathway. Importantly, DHA treatment increased the level of autophagy in HSCs. The inhibition of autophagy by 3-MA dramatically abolished the DHA-induced ferroptosis in HSCs. Mechanistically, the up-regulated m6A modification is essential for the activation of autophagy by DHA through the reduction of fat mass and obesity-associated gene (FTO). Down-regulation of m6A modification by FTO overexpression could impair autophagy and the classical ferroptotic events. Interestingly, the m6A modification of BECN1 mRNA was evidently up-regulated compared with other autophagy-related genes. More importantly, YTHDF1 was identified as a key m6A reader protein for BECN1 mRNA stability, and knockdown of YTHDF1 could prevent DHA-induced HSC ferroptosis. Noteworthy, YTH domain was essential for YTHDF1 to prolong the half-life of BECN1 mRNA in DHA-induced HSC ferroptosis. In mice, DHA treatment alleviated liver fibrosis by triggering HSC ferroptosis. HSC-specific inhibition of m6A modification and autophagy could impair DHA-induced HSC ferroptosis in murine liver fibrosis. Overall, these results provided novel implications to reveal the molecular mechanism of DHA-induced ferroptosis, by which pointed to m6A modification-dependent ferroptosis as a potential target for the treatment of liver fibrosis.

Parallax Attention for Unsupervised Stereo Correspondence Learning.

Stereo image pairs encode 3D scene cues into stereo correspondences between the left and right images. To exploit 3D cues within stereo images, recent CNN based methods commonly use cost volume techniques to capture stereo correspondence over large disparities. However, since disparities can vary significantly for stereo cameras with different baselines, focal lengths and resolutions, the fixed maximum disparity used in cost volume techniques hinders them to handle different stereo image pairs with large disparity variations. In this paper, we propose a generic parallax-attention mechanism (PAM) to capture stereo correspondence regardless of disparity variations. Our PAM integrates epipolar constraints with attention mechanism to calculate feature similarities along the epipolar line to capture stereo correspondence. Based on our PAM, we propose a parallax-attention stereo matching network (PASMnet) and a parallax-attention stereo image super-resolution network (PASSRnet) for stereo matching and stereo image super-resolution tasks. Moreover, we introduce a new and large-scale dataset named Flickr1024 for stereo image super-resolution. Experimental results show that our PAM is generic and can effectively learn stereo correspondence under large disparity variations in an unsupervised manner. Comparative results show that our PASMnet and PASSRnet achieve the state-of-the-art performance.