Search

Lane marking detection is a fundamental but crucial step in intelligent driving systems. It can not only provide relevant road condition information to prevent lane departure but also assist vehicle positioning and forehead car detection. However, lane marking detection faces many challenges, including extreme lighting, missing lane markings, and obstacle obstructions. Recently, deep learning-based algorithms draw much attention in intelligent driving society because of their excellent performance. In this paper, we review deep learning methods for lane marking detection, focusing on their network structures and optimization objectives, the two key determinants of their success. Besides, we summarize existing lane-related datasets, evaluation criteria, and common data processing techniques. We also compare the detection performance and running time of various methods, and conclude with some current challenges and future trends for deep learning-based lane marking detection algorithm.

Facial Attribute Classification (FAC) has attracted increasing attention in computer vision and pattern recognition. However, state-of-the-art FAC methods perform face detection/alignment and FAC independently. The inherent dependencies between these tasks are not fully exploited. In addition, most methods predict all facial attributes using the same CNN network architecture, which ignores the different learning complexities of facial attributes. To address the above problems, we propose a novel deep multi-task multi-label CNN, termed DMM-CNN, for effective FAC. Specifically, DMM-CNN jointly optimizes two closely-related tasks (i.e., facial landmark detection and FAC) to improve the performance of FAC by taking advantage of multi-task learning. To deal with the diverse learning complexities of facial attributes, we divide the attributes into two groups: objective attributes and subjective attributes. Two different network architectures are respectively designed to extract features for two groups of attributes, and a novel dynamic weighting scheme is proposed to automatically assign the loss weight to each facial attribute during training. Furthermore, an adaptive thresholding strategy is developed to effectively alleviate the problem of class imbalance for multi-label learning. Experimental results on the challenging CelebA and LFWA datasets show the superiority of the proposed DMM-CNN method compared with several state-of-the-art FAC methods.

Generative Adversarial Networks (GANs) have shown remarkable success in producing realistic-looking images in the computer vision area. Recently, GAN-based techniques are shown to be promising for spatio-temporal-based applications such as trajectory prediction, events generation and time-series data imputation. While several reviews for GANs in computer vision have been presented, no one has considered addressing the practical applications and challenges relevant to spatio-temporal data. In this paper, we have conducted a comprehensive review of the recent developments of GANs for spatio-temporal data. We summarise the application of popular GAN architectures for spatio-temporal data and the common practices for evaluating the performance of spatio-temporal applications with GANs. Finally, we point out future research directions to benefit researchers in this area., This paper has been accepted by ACM Transactions on Intelligent Systems and Technology (TIST)

To generate a high-spatial-resolution hyperspectral (HHS) image from a high-spatial-resolution multispectral (HMS) image, both spatial information and spectral information should be considered simultaneously if we want to build a more accurate mapping from HMS to HHS. To this end, a spectral and spatial jointed spectral super-resolution method is proposed in this letter using an end-to-end learning strategy for each subspace with the cluster-based multibranch backpropagation neural network (BPNN). More specifically, in addition to the spectra similarity, a modified superpixel segmentation is introduced to jointly take spatial contextual information into account, and a new framework with it is given. Comparisons on the Columbia University Automated Vision Environment (CAVE) data set show that our proposed method outperforms other relative state-of-the-art methods more than 0.3 in the root mean squared error (RMSE) and more than 1.0 in the spectral angle mapper (SAM) index. Especially, an exemplary application is demonstrated using the synchronized observation data collected by the multispectral and hyperspectral sensors mounted on the HJ-1A satellite at the same time.

Video saliency prediction has received increasing attention in the field of computer vision research. How to model the spatio-temporal information in video frames is a key issue for saliency prediction. Most of the existing methods either show a limited temporal memory ability or fail in jointly processing spatial and temporal information. Besides, these methods take each pixel of video frames as input indiscriminately, which may result in incorrect results because of redundant pixels. To solve the above problems, we propose a novel Explicit Cyclic Attention-based Network (ECANet), which is a two-stream encoder-decoder model. Specifically, an explicit cyclic attention mechanism is proposed for temporal modeling and pixel emphasizing, in which an attentional stream is built, whose input is cropped patches based on the model’s previous prediction. For spatio-temporal information aggregation, 3D convolutional neural network is adopted as the backbone of our encoders and decoder. Stacks of video frames and attentional patches from past steps build up the input of our model. Extensive experimental results over three benchmark datasets show that our model achieves competitive performance against state-of-the-art methods.

Different from the object motion blur, the defocus blur is caused by the limitation of the cameras' depth of field. The defocus amount can be characterized by the parameter of point spread function and thus forms a defocus map. In this paper, we propose a new network architecture called Defocus Image Deblurring Auxiliary Learning Net (DID-ANet), which is specifically designed for single image defocus deblurring by using defocus map estimation as auxiliary task to improve the deblurring result. To facilitate the training of the network, we build a novel and large-scale dataset for single image defocus deblurring, which contains the defocus images, the defocus maps and the all-sharp images. To the best of our knowledge, the new dataset is the first large-scale defocus deblurring dataset for training deep networks. Moreover, the experimental results demonstrate that the proposed DID-ANet outperforms the state-of-the-art methods for both tasks of defocus image deblurring and defocus map estimation, both quantitatively and qualitatively. The dataset, code, and model is available on GitHub: https://github.com/xytmhy/DID-ANet-Defocus-Deblurring.

Polarization images encode high resolution microstructural information even at low resolution. We propose a framework combining polarization imaging and traditional microscopy imaging, constructing a dual-modality machine learning framework that is not only accurate but also generalizable and interpretable. We demonstrate the viability of our proposed framework using the cervical intraepithelial neoplasia grading task, providing a polarimetry feature parameter to quantitatively characterize microstructural variations with lesion progression in hematoxylin-eosin-stained pathological sections of cervical precancerous tissues. By taking advantages of polarization imaging techniques and machine learning methods, the model enables interpretable and quantitative diagnosis of cervical precancerous lesion cases with improved sensitivity and accuracy in a low-resolution and wide-field system. The proposed framework applies routine image-analysis technology to identify the macro-structure and segment the target region in H&E-stained pathological images, and then employs emerging polarization method to extract the micro-structure information of the target region, which intends to expand the boundary of the current image-heavy digital pathology, bringing new possibilities for quantitative medical diagnosis.

OBJECTIVE: This study evaluated the prognostic power of serum uric acid (UA) in predicting adverse events in elderly acute coronary syndrome (ACS) patients with diabetes mellitus (DM). METHODS: The analysis involved 718 ACS patients ‍>80 years old whose general clinical data and baseline blood biochemical indicators were collected prospectively from January 2006 to December 2012. These patients were classified into two groups based on DM status, and then followed up after discharge. The Kaplan-Meier method was used for major adverse cardiac event (MACE) rates and all-cause mortality. Multivariate Cox regression was performed to analyze the relationship between UA level and long-term clinical prognosis. Receiver operating characteristic (ROC) curves were analyzed to predict the cutoff value of UA in elderly ACS patients with DM. There were 242 and 476 patients in the DM and non-DM (NDM) groups, respectively, and the follow-up time after discharge was 40‒120 months (median, 63 months; interquartile range, 51‒74 months). RESULTS: The all-cause mortality, cardiac mortality, and MACE rates in both DM and NDM patients were higher than those in the control group (P=0.001). All-cause mortalities, cardiac mortalities, and MACE rates in DM patients with moderate and high UA levels were significantly higher than those in the NDM group (P=0.001). Long-term survival rates decreased significantly with increased UA levels in the ACS groups (P=0.001). UA (odds ratio (OR)=2.106, 95% confidence interval (CI)=1.244‒3.568, P=0.006) was found to be an independent risk factor for all-cause mortality and MACE in elderly ACS patients with DM. The cutoff value of UA was 353.6 μmol/L (sensitivity, 67.4%; specificity, 65.7%). CONCLUSIONS: Serum UA level is a strong independent predictor of long-term all-cause death and MACE in elderly ACS patients with DM.

In deep face recognition, the commonly used softmax loss and its newly proposed variations are not yet sufficiently effective to handle the class imbalance and softmax saturation issues during the training process while extracting discriminative features. In this brief, to address both issues, we propose a class-variant margin (CVM) normalized softmax loss, by introducing a true-class margin and a false-class margin into the cosine space of the angle between the feature vector and the class-weight vector. The true-class margin alleviates the class imbalance problem, and the false-class margin postpones the early individual saturation of softmax. With negligible computational complexity increment during training, the new loss function is easy to implement in the common deep learning frameworks. Comprehensive experiments on the LFW, YTF, and MegaFace protocols demonstrate the effectiveness of the proposed CVM loss function.

Background Although the progress in global health initiatives has improved the availability of primary health care (PHC), unqualified healthcare remains a serious challenge in low- and middle-income countries, where PHC is often underutilized. This study examines factors associated with patients’ healthcare-seeking behaviors in rural Chin—seeking healthcare at village-level PHC providers, at higher-level health facilities, self-medicating, and refraining from seeking medical help. We focus on provider-side factors, including (1) the unobservable quality indicator, (2) the observable quality indicator, and (3) the observable signal indicator. Methods We analyzed 1578 episodes of healthcare-seeking behaviors of patients with diarrhea or cough/runny nose symptom from surveys conducted in July 2017 and January 2018 in 114 villages of the Yunnan province. We investigated the correlation between quality-related factors with patients’ healthcare-seeking behaviors by multinomial logit regression. Results We found that rural patients were insensitive to the unobservable quality of healthcare providers, as measured by standardized clinical vignettes, which might be attributable to the credence nature of PHC. The observable quality indicator, whether the clinician has received full-time junior college formal medical education, was associated with patients’ healthcare choices. Patients, however, were more likely to select healthcare based on the observable signal indicator, which was measured by the availability of medicines. Additionally, the observable signal indicator had no significant association with two quality indicators. Notably, socioeconomically-disadvantaged patients relied more on the village-level PHC, which emphasized the role of PHC in promoting the welfare of rural populations. Conclusions Our study found an inconsistency between objective quality of healthcare provided by providers and subjective quality perceived by patients. Patients could not identify the actual quality of PHC precisely, while they were more likely to make decisions based on the observable signal indicator. Therefore, the quality of PHC should be more observable to patients. This study not only supplements the literature on healthcare-seeking choices by examining four types of behaviors simultaneously but also clarifies rural patients’ perceptions of the quality of PHC for policy decision-making on increasing the utilization of PHC and improving the medical welfare of the vulnerable.

Scene text detection plays a vital role for scene text understanding, but arbitrary-shaped text detection remains a significant challenge. To extract discriminative features, most recent state-of-the-art methods adopt heavy networks, resulting in parameter redundancy and inference inefficiency. For accurate and efficient scene text detection, in this paper we propose a novel recurrent feature refinement network (RFRN). RFRN, as a recurrent segmentation framework, contains a recurrent path augmentation that refines the previous feature maps as inner states, which not only helps improve the segmentation quality, but also fully facilitates the reuse of parameters and low computational cost. During testing, RFRN discards redundant prediction procedures for efficient inference, and achieves a good balance between speed and accuracy of inference. We conduct experiments on four challenging scene text benchmarks, CTW1500, Total-Text, ICDAR2015 and ICDAR2017-MLT, which include curved texts and multi-oriented texts with complex background. The results show that the proposed RFRN achieves competitive performance on detection accuracy while maintaining computational efficiency.

Background Extrahepatic cholangiocarcinoma is a malignancy that originates from bile duct epithelium with an unfavorable prognosis. Epithelial membrane protein 1 was first discovered in 1995, functioning as an oncogene or anti-tumor gene in various cancers. However, the clinical role of epithelial membrane protein 1 extrahepatic cholangiocarcinoma remained unclear. Methods Differentially expressed genes were identified using Gene Ontology and the Kyoto Encyclopedia and Genomes pathway analysis. Out of 183 extrahepatic cholangiocarcinoma patients and 61 healthy controls, the expression level of epithelial membrane protein 1 was detected and compared using reverse transcription-quantitative polymerase chain reaction analysis and western blot assay. Meanwhile, the diagnosis and prognosis of EMP1 in ECCA were measured by receiver operating characteristic and Kaplan–Meier analysis. Finally, the relationship between epithelial membrane protein 1 expression and clinicopathological indexes were compared to further verify the clinical role of epithelial membrane protein 1 in extrahepatic cholangiocarcinoma. Results After analyzing data from GSE76297, GSE89749, and GSE26566GO, we found 1554 down-regulated and 1065 up-regulated genes. Through Gene Ontology and Kyoto Encyclopedia and Genomes analysis, extracellular matrix organization, extracellular structure organization, cholesterol metabolism, interleukin-17 signaling pathway, and vitamin digestion and absorption were significantly enriched and involved in targeted differentially expresses genes. Epithelial membrane protein 1 messenger ribonucleic acid was notably decreased in serum samples from extrahepatic cholangiocarcinoma patients, compared with that in healthy controls. Receiver operating characteristic analysis revealed that the area under the curve of epithelial membrane protein 1 messenger ribonucleic acid for the diagnosis of extrahepatic cholangiocarcinoma was 0.9281 (95% CI = 0.8967–0.9595). Moreover, the correlation analysis presented that epithelial membrane protein 1 expression was negatively correlated with lymph node metastasis, tumour node metastasis stage, cancer antigen 19-9 level, and carcinoembryonic antigen level. Conclusion Aberrant expression of epithelial membrane protein 1 contributed to distinguishing extrahepatic cholangiocarcinoma patients and healthy controls, and a low expression level of epithelial membrane protein 1 indicated an unfavorable prognosis. Hence, epithelial membrane protein 1 was a feasible and credible biomarker for extrahepatic cholangiocarcinoma diagnosis and prognosis, with high accuracy, sensitivity, and specificity.

Traffic sign detection and recognition in the wild is a challenging task. Existing techniques are often incapable of detecting small or occluded traffic signs because of the scale variation and context loss, which causes semantic gaps between multiple scales. We propose a new traffic sign detection network (TSingNet), which learns scale-aware and context-rich features to effectively detect and recognize small and occluded traffic signs in the wild. Specifically, TSingNet first constructs an attention-driven bilateral feature pyramid network, which draws on both bottom-up and top-down subnets to dually circulate low-, mid-, and high-level foreground semantics in scale self-attention learning. This is to learn scale-aware foreground features and thus narrow down the semantic gaps between multiple scales. An adaptive receptive field fusion block with variable dilation rates is then introduced to exploit context-rich representation and suppress the influence of occlusion at each scale. TSingNet is end-to-end trainable by joint minimization of the scale-aware loss and multi-branch fusion losses, this adds a few parameters but significantly improves the detection performance. In extensive experiments with three challenging traffic sign datasets (TT100K, STSD and DFG), TSingNet outperformed state-of-the-art methods for traffic sign detection and recognition in the wild.

Video action recognition is a vital area of computer vision. By adding temporal dimension into convolution structure, 3D convolution neural network owns the capacity to extract spatio-temporal features from videos. However, due to computing constraints, it is hard to input the whole video into the convolution network at one time, resulting in a limited temporal receptive field of the network. To address this issue, we propose a novel 3D temporal dilation convolution (3D-TDC) framework, to extract spatio-temporal features of actions from videos. First, we deploy the 3D temporal dilation convolution as the shallow temporal compression layer, enabling an effective capture of spatio-temporal information in a larger time domain with the reduced computational load. Then, an action recognition framework is constructed by integrating two networks with different temporal receptive fields to balance the long-short time difference. We conduct extensive experiments on three widely-used public datasets (UCF-101, HMDB-51, and Kinetics-400) for performance evaluation, and the experimental results demonstrate the effectiveness of our proposed framework in video action recognition with low computational load.

In recent years, deep learning-based person re-identification (Re-ID) methods have made significant progress. However, the performance of these methods substantially decreases when dealing with occlusion, which is ubiquitous in realistic scenarios. In this article, we propose a novel semantic-aware occlusion-robust network (SORN) that effectively exploits the intrinsic relationship between the tasks of person Re-ID and semantic segmentation for occluded person Re-ID. Specifically, the SORN is composed of three branches, including a local branch, a global branch, and a semantic branch. In particular, the local branch extracts part-based local features, and the global branch leverages a novel spatial-patch contrastive loss (SPC) to extract occlusion-robust global features. Meanwhile, the semantic branch generates a foreground-background mask for a pedestrian image, which indicates the non-occluded areas of the human body. The three branches are jointly trained in a unified multi-task learning network. Finally, pedestrian matching is performed based on the local features extracted from the non-occluded areas and the global features extracted from the whole pedestrian image. Extensive experimental results on a large-scale occluded person Re-ID dataset (i.e., Occluded-DukeMTMC) and two partial person Re-ID datasets (i.e., Partial-REID and Partial-iLIDS) show the superiority of the proposed method compared with several state-of-the-art methods for occluded and partial person Re-ID. We also demonstrate the effectiveness of the proposed method on two general person Re-ID datasets (i.e., Market-1501 and DukeMTMC-reID).

Low-frequency noise characteristics of ZnO nanowire field-effect transistors (FETs) are presented at temperatures down to 10 K. The carrier number fluctuation (CNF) model and Hooge’s model are used to analyze the low-frequency noise in the studied devices. From 293 down to 200 K, the normalized noise power spectrum density (PSD) is proportional to $({g}_{m}/{I}_{d}{)}^{{2}}$ due to CNF caused by the trapping and detrapping process between the channel carriers and oxide traps near the semiconductor and the dielectrics interface. In this case, the low-frequency noise is induced by flat-band voltage fluctuation. The density of the traps in the gate dielectrics near the Fermi level ranges from 9.4 $\times \,\,10^{{10}}$ to 3.8 $\times \,\,10^{{10}}$ cm $^{-{2}}$ eV $^{-{1}}$ from 293 and 200 K. At lower temperatures, from 150 down to 100 K, the noise mechanism is mobility fluctuation, attributed to electron and space-charge scattering due to the hoping process and slow motions of electrons. The extracted Hooge’s parameter in this temperature ranges in the order of $10^{-{3}}$ to $10^{-{2}}$ . In the temperature range from 30 down to 10 K, the low-frequency noise is dominated by both CNF and mobility fluctuation caused by Coulomb scattering between oxide traps and electrons.

Hyperuricemia is associated with cardiovascular mortality, but the association of the age at hyperuricemia onset with cardiovascular disease (CVD) and mortality is still unclear. The purpose of this study was to examine the associations of hyperuricemia onset age with CVD and all-cause mortality. A total of 82,219 participants free of hyperuricemia and CVD from 2006 to 2015 in the Kailuan study were included. The analysis cohort comprised 18,311 new-onset hyperuricemia patients and controls matched for age and sex from the general population. Adjusted associations were estimated using Cox models for CVD and all-cause mortality across a range of ages. There were 1,021 incident cases of CVD (including 215 myocardial infarctions, 814 strokes) and 1459 deaths during an average of 5.2 years of follow-up. Patients with hyperuricemia onset at an age

Human fingerprints are detailed and nearly unique markers of human identity. Such a unique and stable fingerprint is also left on each acquired image. It can reveal how an image was degraded during the image acquisition procedure and thus is closely related to the quality of an image. In this work, we propose a new no-reference image quality assessment (NR-IQA) approach called domain-aware IQA (DA-IQA), which for the first time introduces the concept of domain fingerprint to the NR-IQA field. The domain fingerprint of an image is learned from image collections of different degradations and then used as the unique characteristics to identify the degradation sources and assess the quality of the image. To this end, we design a new domain-aware architecture, which enables simultaneous determination of both the distortion sources and the quality of an image. With the distortion in an image better characterized, the image quality can be more accurately assessed, as verified by extensive experiments, which show that the proposed DA-IQA performs better than almost all the compared state-of-the-art NR-IQA methods., accepted by IEEE Transactions on Circuits and Systems for Video Technology (TCSVT)

We propose an AgGeSe/Al2O3/Pt selector with great potential in a dense memory array. In terms of on-current drive, selectivity, and uniformity, devices with the mixed AgGeSe layer outperform the Ag/Al2O3/Pt device. The introduction of GeSe not only blocks the diffusion of Ag, but also facilitates the backflow of Ag to the active electrode, therefore increasing the on-current. The ultra-high on-current drive (3 mA and 12.7 MA/cm2), high selectivity (1010), small variation, high thermal stability, and steep switching slope (0.27 mV/dec) realized in our AgGeSe/Al2O3/Pt device render it a promising candidate for selector application.

With artificial intelligence technology being advanced by leaps and bounds, intelligent driving has attracted a huge amount of attention recently in research and development. In intelligent driving, lane line detection is a fundamental but challenging task particularly under complex road conditions. In this paper, we propose a simple yet appealing network called Ripple Lane Line Detection Network (RiLLD-Net), to exploit quick connections and gradient maps for effective learning of lane line features. RiLLD-Net can handle most common scenes of lane line detection. Then, in order to address challenging scenarios such as occluded or complex lane lines, we propose a more powerful network called Ripple-GAN, by integrating RiLLD-Net, confrontation training of Wasserstein generative adversarial networks, and multi-target semantic segmentation. Experiments show that, especially for complex or obscured lane lines, Ripple-GAN can produce a superior detection performance to other state-of-the-art methods.

Automatic speech recognition (ASR) is a crucial technology for man-machine interaction. End-to-end models have been studied recently in deep learning for ASR. However, these models are not suitable for the practical application of ASR due to their large model sizes and computation costs. To address this issue, we propose a novel mutual-learning sequence-level knowledge distillation framework enjoying distinct student structures for ASR. Trained mutually and simultaneously, each student learns not only from the pre-trained teacher but also from its distinct peers, which can improve the generalization capability of the whole network, through making up for the insufficiency of each student and bridging the gap between each student and the teacher. Extensive experiments on the TIMIT and large LibriSpeech corpuses show that, compared with the state-of-the-art methods, the proposed method achieves an excellent balance between recognition accuracy and model compression.

Weakly supervised semantic segmentation has been attracting increasing attention as it can alleviate the need for expensive pixel-level annotations through the use of image-level labels. Relevant methods mainly rely on the implicit object localization ability of convolutional neural networks (CNNs). However, generated object attention maps remain mostly small and incomplete. In this paper, we propose an Atrous Convolutional Feature Network (ACFN) to generate dense object attention maps. This is achieved by enhancing the context representation of image classification CNNs. More specifically, cascaded atrous convolutions are used in the middle layers to retain sufficient spatial details, and pyramidal atrous convolutions are used in the last convolutional layers to provide multi-scale context information for the extraction of object attention maps. Moreover, we propose an attentive fusion strategy to adaptively fuse the multi-scale features. Our method shows improvements over existing methods on both the PASCAL VOC 2012 and MS COCO datasets, achieving state-of-the-art performance.

Exploiting both RGB (2D appearance) and Depth (3D geometry) information can improve the performance of semantic segmentation. However, due to the inherent difference between the RGB and Depth information, it remains a challenging problem in how to integrate RGB-D features effectively. In this letter, to address this issue, we propose a Non-local Aggregation Network (NANet), with a well-designed Multi-modality Non-local Aggregation Module (MNAM), to better exploit the non-local context of RGB-D features at multi-stage. Compared with most existing RGB-D semantic segmentation schemes, which only exploit local RGB-D features, the MNAM enables the aggregation of non-local RGB-D information along both spatial and channel dimensions. The proposed NANet achieves comparable performances with state-of-the-art methods on popular RGB-D benchmarks, NYUDv2 and SUN-RGBD.

Background Wrist tuberculosis is a rare disease, which is easy to be misdiagnosed, leading to delayed treatment and poor prognosis. In this study, the clinical manifestations, diagnosis, treatment, and prognosis of 18 cases of wrist tuberculosis were analyzed retrospectively. Methods A retrospective study was conducted, investigating tuberculosis of the wrist, diagnosed in 18 patients from August 2013 to November 2018. Puncture biopsy confirmed the diagnosis. The study includes 11 males and 7 females, and 8 left and 10 right wrists. The average age was 53.5 ± 18.3 years and ranged from 15 to 81 years. The disease course was 1 to 42 months, with an average of 15.1 ± 11.3 months. Eighteen patients were underwent surgery and chemotherapy, 3 patients with severe bone defects were treated with wrist fusion, and 15 patients were underwent focus removal. The Gartland and Werley score, DASH score, the range of motion (ROM), grip strength, and imaging examinations were used to evaluate the postoperative recovery of the patients. Results Eighteen patients were followed up for 15 to 77 months, with an average follow up of 39.7 ± 15.3 months. The ESR and CRP levels were normal for all patients after chemotherapy. No recurrence of tuberculosis was observed in any of the patients. Among the 15 focus removals, the Gartland and Werley scores at admission, two weeks of chemotherapy, 1 month after surgery, and the last follow-up were 21.73 ± 4.33, 18.60 ± 3.16,11.27 ± 2.79, and 5.07 ± 2.28, respectively; and the DASH scores were 45.87 ± 5.58, 39.47 ± 4.72, 22.67 ± 6.54, and 6.73 ± 2.94, respectively. The range of motion (ROM) of the wrist and grip strength improved significantly when compared to those at admission. Among the three cases of wrist fusion, 2 were fixed with a steel plate and the fixation position of wrist joint was good. One case was fixed with Kirschner wire and resulted in a slightly deformed wrist joint. Conclusion For patients with wrist tuberculosis, early diagnosis, preoperative and postoperative chemotherapy, thorough focus removal, and appropriate fixation of the affected limb can help restore the function of the affected wrist, reduce the recurrence rate, and improve the quality of life.

Silicone rubber elastomers are broadly used in various fields, where the three-dimensional (3D) printing of silicone rubber elastomers is important for the free construction of complex structures. Herein, a series of polydimethylsiloxane/polytetrafluoroethylene composite inks for direct-ink-writing 3D printing are developed. The inks are prepared by directly mixing a silicone rubber liquid precursor with polytetrafluoroethylene micropowder. The polytetrafluoroethylene micropowder serves as a thixotropic agent to regulate the rheological properties of the polydimethylsiloxane precursor to fulfill the requirement of 3D printing and endow the composite material with high electron affinity. The printed polydimethylsiloxane/polytetrafluoroethylene composite elastomer exhibits excellent elasticity and cyclic stability. A high-performance triboelectric nanogenerator is constructed with the 3D-printed polydimethylsiloxane/polytetrafluoroethylene composite as the triboelectric layer and elastic structure. This work establishes a new method of 3D printing polydimethylsiloxane-based elastomers and thus provides a new technique for constructing complex structures in flexible devices.

In this paper, we propose a new loss function, termed the equalized margin (EqM) loss, which is designed to make both intra-class scopes and inter-class margins similar over all classes, such that all the classes can be evenly distributed on the hypersphere of the feature space. The EqM loss controls both the lower limit of intra-class similarity by exploiting hard-sample mining and the upper limit of inter-class similarity by assuring equalized margins. Therefore, using the EqM loss, we can not only obtain more discriminative features, but also overcome the negative impacts from the data imbalance on the inter-class margins. We also observe that the EqM loss is stable with the variation of the scale in normalized Softmax. Furthermore, by conducting extensive experiments on LFW, YTF, CFP, MegaFace and IJB-B, we are able to verify the effectiveness and superiority of the EqM loss, compared with other state-of-the-art loss functions for face recognition.

Multi-focus image fusion, a technique to generate an all-in-focus image from two or more partially-focused source images, can benefit many computer vision tasks. However, currently there is no large and realistic dataset to perform convincing evaluation and comparison of algorithms in multi-focus image fusion. Moreover, it is difficult to train a deep neural network for multi-focus image fusion without a suitable dataset. In this letter, we introduce a large and realistic multi-focus dataset called Real-MFF, which contains 710 pairs of source images with corresponding ground truth images. The dataset is generated by light field images, and both the source images and the ground truth images are realistic. To serve as both a well-established benchmark for existing multi-focus image fusion algorithms and an appropriate training dataset for future development of deep-learning-based methods, the dataset contains a variety of scenes, including buildings, plants, humans, shopping malls, squares and so on. We also evaluate 10 typical multi-focus algorithms on this dataset for the purpose of illustration.

The initial aim of environmental epidemiology is to estimate the causal effects of environmental exposures on health outcomes. However, due to lack of enough covariates in most environmental data sets, current methods without enough adjustments for confounders inevitably lead to residual confounding. We propose a negative-control exposure based on a time-series studies (NCE-TS) model to effectively eliminate unobserved confounders using an after-outcome exposure as a negative-control exposure. We show that the causal effect is identifiable and can be estimated by the NCE-TS for continuous and categorical outcomes. Simulation studies indicate unbiased estimation by the NCE-TS model. The potential of NCE-TS is illustrated by 2 challenging applications: We found that living in areas with higher levels of surrounding greenness over 6 months was associated with less risk of stroke-specific mortality, based on the Shandong Ecological Health Cohort during January 1, 2010, to December 31, 2018. In addition, we found that the widely established negative association between temperature and cancer risks was actually caused by numbers of unobserved confounders, according to the Global Open Database from 2003–2012. The proposed NCE-TS model is implemented in an R package (R Foundation for Statistical Computing, Vienna, Austria) called NCETS, freely available on GitHub.

Seismic facies analysis is to study the sedimentary environment of stratigraphic sequence and provides an important basis for reservoir prediction. Most of the existing analysis methods have low efficiency and heavily rely on manual experience, and therefore, it is difficult to interpret increasingly complex seismic data. Deep learning techniques can help to solve these problems and achieve automatic seismic facies classification. We regard seismic facies classification as a target segmentation problem and propose new method and training strategies. Our workflow primarily involves four sections. First, we process the manually annotated labels and seismic data with mirroring and cropping operations to ensure that network can accept input with arbitrary size and the model training is not limited to GPU memory. Second, data augmentation is applied to automatically generate massive training samples from the processed data. Third, we build two independent networks based on encoder–decoder architecture: one identifies all seismic facies simultaneously, and the other identifies single seismic facies in each model. However, both the results of the two networks have some drawbacks. Fourth, to overcome these drawbacks, we propose an ensemble learning method to get optimized model and test it on 3-D seismic data. The testing results manifest that the proposed method can improve the predictive ability of model, accurately describe the seismic facies, and can be applicable to entire seismic data volume.

Lead halide perovskites have shown great X-ray scintillation properties. One big challenge is to find nontoxic replacements for practical applications. Moreover, the scintillator should contain les...

The relations of hypertension onset age with cardiovascular diseases (CVD) and all-cause mortality remain inconclusive.This study sought to examine the associations of hypertension onset age with CVD and all-cause mortality.This prospective study included 71,245 participants free of hypertension and CVD in the first survey (July 2006 to October 2007) of the Kailuan study, a prospective cohort study in Tangshan, China. All participants were followed biennially until December 31, 2017. A total of 20,221 new-onset hypertension cases were identified during follow-up. We randomly selected 1 control participant for each new-onset hypertensive participant, matching for age (±1 year) and sex, and included 19,887 case-control pairs. We used weighted Cox regression models to calculate the average hazard ratios of incident CVD and all-cause mortality across the age groups.During an average follow-up of 6.5 years, we identified 1,672 incident CVD cases and 2,008 deaths. After multivariate adjustment, with the increase in hypertension onset age, the hazards of outcomes were gradually attenuated. The average hazard ratio (95% confidence interval) of CVD and all-cause mortality were 2.26 (1.19 to 4.30) and 2.59 (1.32 to 5.07) for the hypertension onset age 45 years old group, 1.62 (1.24 to 2.12) and 2.12 (1.55 to 2.90) for the 45- to 54-year age group, 1.42 (1.12 to 1.79) and 1.30 (1.03 to 1.62) for the 55- to 64-year age group, and 1.33 (1.04 to 1.69) and 1.29 (1.11 to 1.51) for the ≥65-year age group, respectively (p for interaction = 0.38 for CVD and 0.01 for death).Hypertension was associated with a higher risk for CVD and all-cause mortality, and the associations were stronger with a younger age of onset.

Metal halide perovskites have recently been reported as excellent scintillators for X-ray detection. However, perovskite based scintillators are susceptible to moisture and oxygen atmosphere, such as the water solubility of CsPbBr3, and oxidation vulnerability of Sn2+, Cu+. The traditional metal halide scintillators (NaI: Tl, LaBr3, etc.) are also severely restricted by their high hygroscopicity. Here we report a new kind of lead free perovskite with excellent water and radiation stability, Rb2Sn1-x Te x Cl6. The equivalent doping of Te could break the in-phase bonding interaction between neighboring octahedra in Rb2SnCl6, and thus decrease the electron and hole dimensionality. The optimized Te content of 5% resulted in high photoluminescence quantum yield of 92.4%, and low X-ray detection limit of 0.7 µGyair s−1. The photoluminescence and radioluminescence could be maintained without any loss when immersing in water or after 480,000 Gy radiations, outperforming previous perovskite and traditional metal halides scintillators.

Class-imbalance extent metrics measure how imbalanced the data are. In pattern classification, it is usually expected that the higher the imbalance extent, the worse the classification performance, and thus an appropriate imbalance extent metric should show a negative correlation with the classification performance. Existing metrics, such as the popular imbalance ratio (IR), only consider the effect of the sample sizes of different classes. However, we note that the dimensionality of imbalanced data also affects the classification performance. Datasets with the same IR can present distinct classification performances when their dimensionalities are different, making IR suboptimal to reflect the imbalance extent for classification. We also observe that the classification performance becomes better with more discriminative features. Inspired by these observations, we propose a new imbalance extent metric, the adjusted IR, by adding a penalty term of the number of discriminative features that is effectively determined by the Pearson correlation test. The adjusted IR adaptively revises the IR when the number of discriminative features varies. The empirical studies demonstrate the effectiveness of the adjusted IR, in terms of its better negative correlation with the classification performance.

Background. The association between dietary sodium intake and blood pressure variability (BPV) in hypertensive patients remains unclear. The objective of this study was to demonstrate whether dietary sodium intake is a predictor of elevated BPV in Chinese patients with hypertension. Methods. A total of 235 patients with essential hypertension were enrolled in the Department of Cardiology, Chinese People's Liberation Army (PLA) General Hospital in 2018 to 2019, all of whom underwent 24-h ambulatory blood pressure monitoring. BPV was calculated as the standard deviation (SD), coefficient of variation (CV), variation independent of mean (VIM) of blood pressure measurements, respectively, and divided into diurnal systolic BPV (SBPV), diurnal diastolic BPV (DBPV), nocturnal SBPV, and nocturnal DBPV. 24-h urine samples were collected to measure 24-h urine sodium excretion, which represents dietary sodium intake. The relationship between dietary sodium intake and BPV was analyzed by using Spearman correlations and multiple linear regression analysis. Results. Nocturnal SBPV-SD, CV, VIM, and nocturnal DBPV-SD in the high urine sodium excretion group were significantly higher than those in the medium and low urine sodium excretion groups, whereas diurnal SBPV-SD, CV, VIM, diurnal DBPV-SD, CV, VIM, and nocturnal DBPV-CV, VIM were not. Using the Spearman correlation analysis, we found a linear correlation between 24-h urine sodium excretion and nocturnal SBPV-SD, CV, VIM (SD, r = 0.22, P = 0.001; CV, r = 0.17, P = 0.009; VIM, r = 0.16, P = 0.020), nocturnal DBPV-SD (r = 0.21, P = 0.001), respectively. After further adjusting for confounding factors by multiple linear regression, the positive correlations remained between 24-h urine sodium excretion and nocturnal SBPV-SD, CV, VIM (SD, β = 0.224, P

Atherosclerosis (AS), known as the chronic inflammatory disease, results from the dysfunction of vascular endothelial cells (VECs). Transforming growth factor-β1 (TGF-β1) has been reported to be induced by oxidized low-density lipoprotein (ox-LDL) and contribute to AS-related vascular endothelial cell damage. This work planned to study the mechanism of TGF-β1 in vascular endothelial cell damage. We found that TGF-β1 was activated by ox-LDL in human umbilical vascular endothelial cells (HUVECs). Silence of TGF-β1 reversed the inductive effect of ox-LDL on apoptosis and inflammatory response of HUVECs. Mechanistically, microRNA-4286 (miR-4286) targeted and inhibited TGF-β1 to inhibit Smad3, and Smad3 bound to the promoter of miR-4286 to repress its transcription. Rescue assays indicated that miR-4286 ameliorated the ox-LDL-induced apoptosis and inflammatory response through inhibiting TGF-β1. In conclusion, our study first demonstrated that miR-4286/TGF-β1/Smad3-negative feedback loop ameliorated vascular endothelial cell damage by attenuating apoptosis and inflammatory response, providing new thoughts for promoting the treatment of AS.

A common means of incorporating non-verifiable performance measures in compensation contracts is via bonus pools. We study a principal-multi-agent relational contracting model in which the optimal contract resembles a bonus pool. It specifies a minimum joint bonus floor the principal is required to pay out to the agents and gives the principal discretion to use non-verifiable performance measures to both increase the size of the pool and to allocate the pool to the agents. The joint bonus floor is useful because of its role in motivating the agents to mutually monitor each other (team incentives). In an extension section, we introduce a verifiable team performance measure, in part to establish the robustness of our results. In this case, the optimal contract either ignores the team measure completely or uses it to create a conditional bonus floor (a floor only when the team measure is high). There is a minimum precision requirement on the team measure that has to be satisfied before the team measure is used to motivate mutual monitoring.

In this paper, we propose a method to generate a high-order Bessel orbital angular momentum (OAM) vortex beam by using a single-layer reflective metasurface, which integrates the advantages of small size, high efficiency, insensitive polarization and stable incident angles. An offset-fed horn configuration is adopted to overcome the feed-blockage effect, and a subwavelength unit cell with rotational symmetry is designed to cover the reflection phase variation range of 360° in different incident angles. The metasurface is simulated, fabricated and measured at the center frequency of 10 GHz, and the results validate that a second-order Bessel OAM vortex beam can be generated effectively. By comparing the field distributions between the Bessel OAM vortex beam and the conventional OAM vortex beam, we find that the generated Bessel OAM vortex beam is obviously more convergent and has more stable field distributions. The single-layer reflective metasurface is simple and flexible to shape the wavefront to be a Bessel OAM vortex wave, which has the potential to be used in a wide range of applications.

Two-dimensional (2D) materials with both ferroelasticity and negative Poisson's ratios have attracted intensive interest, but it is very rare to have both ferroelasticity and negative Poisson's ratios in a single material. Directional positive and negative Poisson's ratios in a switchable ferroelastic dielectric may enable non-destructive readout in ferroelastic data storage. Herein, we propose 14 kinds of stable 2D semiconductors: AB monolayers (A = Sc, Y, La; B = N, P, As, Sb, Bi) based on first-principles calculations. The band gaps of AB monolayers cover a wide range from 0.69 eV to 2.15 eV. Mechanical analysis reveals that these materials are flexible and 12 of 14 are predicted to possess an in-plane negative Poisson's ratio (NPR). Moreover, 10 of these 14 systems possess an out-of-plane NPR. More encouragingly, all AB monolayers are identified as 2D ferroelastic materials with reversible strains of around 5.94% to 20.30%. The ferroelastic switching barriers, mechanical properties and electronic structures of these materials are discussed in detail. Such outstanding properties make the AB monolayers very promising as switchable anisotropic 2D materials for nanoelectronics and micromechanical applications.

Two-dimensional (2D) organic–inorganic lead halide perovskites have attracted intensive attention for optoelectronic applications. Principally, strain-induced octahedral distortion in these materials may open up new application scope through modulation of carrier transport. However, they have never been explored as strain sensors. Moreover, traditional strain sensors are troubled by their limited sensitivity at a low strain ratio (

The hybrid perovskite CH3NH3PbX3 (X = Cl, Br, I) is a promising material for developing novel optoelectronic devices. Due to its intrinsic non-layered crystal structure, it remains challenging to synthesize two-dimensional (2D) single-crystalline CH3NH3PbX3 with nanoscale thickness. Here, we report a bottom-up approach to fabricate large CH3NH3PbX3 2D crystals via liquid-phase growth on a mica substrate. The strong potassium-halogen interactions at the perovskite/mica interface decrease the interface energy, driving the striking in-plane growth of the perovskite. The grown 2D CH3NH3PbBr3 crystal was characterized as 8 nm in thickness and hundreds of micrometers in lateral size. Weak exciton binding energy was crucial for improving the photoelectric performance of 2D CH3NH3PbBr3. A visible-light photodetector with a metal/insulator/perovskite configuration was finally achieved with a photoresponsivity of 126 A W−1 and a bandwidth exceeding 80 kHz. Our work proves that the liquid-phase growth on mica is a controllable method to grow 2D hybrid CH3NH3PbX3 perovskites, which can facilitate both device applications and fundamental investigations.

Over the past few years, deep convolutional neural network-based methods have made great progress in semantic segmentation of street scenes. Some recent methods align feature maps to alleviate the semantic gap between them and achieve high segmentation accuracy. However, they usually adopt the feature alignment modules with the same network configuration in the decoder and thus ignore the different roles of stages of the decoder during feature aggregation, leading to a complex decoder structure. Such a manner greatly affects the inference speed. In this paper, we present a novel Stage-aware Feature Alignment Network (SFANet) based on the encoder-decoder structure for real-time semantic segmentation of street scenes. Specifically, a Stage-aware Feature Alignment module (SFA) is proposed to align and aggregate two adjacent levels of feature maps effectively. In the SFA, by taking into account the unique role of each stage in the decoder, a novel stage-aware Feature Enhancement Block (FEB) is designed to enhance spatial details and contextual information of feature maps from the encoder. In this way, we are able to address the misalignment problem with a very simple and efficient multi-branch decoder structure. Moreover, an auxiliary training strategy is developed to explicitly alleviate the multi-scale object problem without bringing additional computational costs during the inference phase. Experimental results show that the proposed SFANet exhibits a good balance between accuracy and speed for real-time semantic segmentation of street scenes. In particular, based on ResNet-18, SFANet respectively obtains 78.1% and 74.7% mean of class-wise Intersection-over-Union (mIoU) at inference speeds of 37 FPS and 96 FPS on the challenging Cityscapes and CamVid test datasets by using only a single GTX 1080Ti GPU.

Uric acid(UA) is related with cardiovascular disease, but the association of UA variability with all-cause mortality is rarely known. This study aimed to investigate the relationship between UA variability and all-cause mortality in Kailuan cohort study in northern China. Cohort study. Kailuan community hospitals in Tangshan, Hebei province, Northern China. A total of 55717 participants from Kailuan Study were enrolled, and our study followed up biennially from 2006 to 2010. Clinical records of the participants enrolled were analyzed. UA variation independent of mean (UAVIM) values were calculated and all the participants were quartile grouped into four groups as: Q1(UAVIM

This paper proposes a dual-supervised uncertainty inference (DS-UI) framework for improving Bayesian estimation-based UI in DNN-based image recognition. In the DS-UI, we combine the classifier of a DNN, i.e. , the last fully-connected (FC) layer, with a mixture of Gaussian mixture models (MoGMM) to obtain an MoGMM-FC layer. Unlike existing UI methods for DNNs, which only calculate the means or modes of the DNN outputs’ distributions, the proposed MoGMM-FC layer acts as a probabilistic interpreter for the features that are inputs of the classifier to directly calculate the probabilities of them for the DS-UI. In addition, we propose a dual-supervised stochastic gradient-based variational Bayes (DS-SGVB) algorithm for the MoGMM-FC layer optimization. Unlike conventional SGVB and optimization algorithms in other UI methods, the DS-SGVB not only models the samples in the specific class for each Gaussian mixture model (GMM) in the MoGMM, but also considers the negative samples from other classes for the GMM to reduce the intra-class distances and enlarge the inter-class margins simultaneously for enhancing the learning ability of the MoGMM-FC layer in the DS-UI. Experimental results show the DS-UI outperforms the state-of-the-art UI methods in misclassification detection. We further evaluate the DS-UI in open-set out-of-domain/-distribution detection and find statistically significant improvements. Visualizations of the feature spaces demonstrate the superiority of the DS-UI. Codes are available at https://github.com/PRIS-CV/DS-UI .

One of the main challenges in facial expression recognition (FER) is to address the disturbance caused by various disturbing factors, including common ones (such as identity, pose, and illumination) and potential ones (such as hairstyle, accessory, and occlusion). Recently, a number of FER methods have been developed to explicitly or implicitly alleviate the disturbance involved in facial images. However, these methods either consider only a few common disturbing factors or neglect the prior information of these disturbing factors, thus resulting in inferior recognition performance. In this paper, we propose a novel Dual-branch Disturbance Disentangling Network (D3Net), mainly consisting of an expression branch and a disturbance branch, to perform effective FER. In the disturbance branch, a label-aware sub-branch (LAS) and a label-free sub-branch (LFS) are elaborately designed to cope with different types of disturbing factors. On the one hand, LAS explicitly captures the disturbance due to some common disturbing factors by transfer learning on a pretrained model. On the other hand, LFS implicitly encodes the information of potential disturbing factors in an unsupervised manner. In particular, we introduce an Indian buffet process (IBP) prior to model the distribution of potential disturbing factors in LFS. Moreover, we leverage adversarial training to increase the differences between disturbance features and expression features, thereby enhancing the disentanglement of disturbing factors. By disentangling the disturbance from facial images, we are able to extract discriminative expression features. Extensive experiments demonstrate that our proposed method performs favorably against several state-of-the-art FER methods on both in-the-lab and in-the-wild databases.

Video super-resolution (VSR) and video frame interpolation (VFI) are inter-dependent for enhancing videos of low resolution and low frame rate. However, most studies treat VSR and temporal VFI as independent tasks. In this work, we design a spatial-temporal super-resolution network based on exploring the interaction between VSR and VFI. The main idea is to improve the middle frame of VFI by the super-resolution (SR) frames and feature maps from VSR. In the meantime, VFI also provides extra information for VSR and thus, through interacting, the SR of consecutive frames of the original video can also be improved by the feedback from the generated middle frame. Drawing on this, our approach leverages a simple interaction of VSR and VFI and achieves state-of-the-art performance on various datasets. Due to such a simple strategy, our approach is universally applicable to any existing VSR or VFI networks for effectively improving their video enhancement performance.

PurposeSpiritual well-being had protective effect on quality of life in cancer, due to the cultural, regional and custom differences,it was rarely been discussed between cancer and chronic diseases in Chongqing,China. We aimed at comparing the level of spirituality in two groups, and discussing its factors of subjects with cancer at county regions.MethodsA cross-sectional questionnaire survey was distributed to 630 inpatients who received treatment between January and December 2020 in Chongqing University Three Gorges Hospital.In addition to basic demographic data,spirituality was measured using the Chinese version of Functional Assessment of Chronic Illness Therapy-Spiritual Well-Being (FACIT-Sp-12). The mean,standard deviation,independent t-tests, ANOVA and multiple regression were used for statistical description and analysis.ResultsSignificant differences were found between cancer and chronic diseases in total scores of FACIT-Sp-12 and each domain(PP

Deep neural networks have recently shown excellent performance on numerous image classification tasks. These networks often need to estimate a large number of parameters and require much training data. When the amount of training data is small, however, a network with high flexibility quickly overfits the training data, resulting in a large model variance and poor generalization. To address this problem, we propose a new, simple yet effective ensemble method called InterBoost for small-sample image classification. In the training phase, InterBoost first randomly generates two sets of complementary weights for training data, which are used for separately training two base networks of the same structure, and then the two sets of complementary weights are updated for refining the training of the networks through interaction between the two base networks previously trained. This interactive training process continues iteratively until a stop criterion is met. In the testing phase, the outputs of the two networks are combined to obtain one final score for classification. Experimental results on four small-sample datasets, UIUC-Sports, LabelMe, 15Scenes and Caltech101, demonstrate that the proposed ensemble method outperforms existing ones. Moreover, results from the Wilcoxon signed-rank tests show that our method is statistically significantly better than the methods compared. Detailed analysis is also provided for an in-depth understanding of the proposed method.

Nitrogen pollution is a severe problem in the Songhua River Basin (SHR) in China. Samples were collected from 36 sections of the SHR during the high, low, and flat seasons of the river, and the main sources of nitrogen in the water were qualitatively analyzed with isotope data for nitrogen and oxygen of nitrate. The contribution rates of each major pollution source were quantitatively analyzed using the Iso Source mass balance model. The results from these experiments indicate that the values for δ15N-NO3 and δ18O-NO3 in the flat flow season range from 1.52‰ to 14.55‰ and −14.26‰ to 2.03‰, respectively. The values for δ15N-NO3 and δ18O-NO3 in the low flow season range from 6.66‰ to 15.46‰ and −5.82‰ to 65.70‰, respectively. In the low flow season, nitrogen comes from the input of domestic and manure sewage (53%) and soil organic N (45%). The values of δ15N-NO3 and δ18O-NO3 in the high flow season range from 2.07‰ to 14.24‰ and −3.99‰ to 8.03‰, respectively. In the high flow season, nitrogen comes from soil organic nitrogen (41%), domestic and manure sewage (32%), and nitrogen fertilizer (27%), which are the main sources of nitrogen pollution in the SHR. The conclusions from the qualitative and quantitative analysis of nitrogen sources in the SHR can provide a scientific basis for the source control and treatment of nitrogen pollution.

Background The study was performed to evaluate the feasibility of utilizing small intestinal submucosa (SIS) scaffolds seeded with adipose-derived mesenchymal stem cells (ADMSCs) for engineered tendon repairing rat Achilles tendon defects and to compare the effects of preconditioning treatments (hypoxic vs. normoxic) on the tendon healing. Methods Fifty SD rats were randomized into five groups. Group A received sham operation (blank control). In other groups, the Achilles tendon was resected and filled with the original tendon (Group B, autograft), cell-free SIS (Group C), or SIS seeded with ADMSCs preconditioned under normoxic conditions (Group D) or hypoxic conditions (Group E). Samples were collected 4 weeks after operation and analyzed by histology, immunohistochemistry, and tensile testing. Results Histologically, compared with Groups C and D, Group E showed a significant improvement in extracellular matrix production and a higher compactness of collagen fibers. Group E also exhibited a significantly higher peak tensile load than Groups D and C. Additionally, Group D had a significantly higher peak load than Group C. Immunohistochemically, Group E exhibited a significantly higher percentage of MKX + cells than Group D. The proportion of ADMSCs simultaneously positive for both MKX and CM-Dil observed from Group E was also greater than that in Group D. Conclusions In this animal model, the engineered tendon grafts created by seeding ADMSCs on SIS were superior to cell-free SIS. The hypoxic precondition further improved the expression of tendon-related genes in the seeded cells and increased the rupture load after grafting in the Achilles tendon defects.