Conservation planning on China's borders with Myanmar, Laos, and Vietnam.

Transboundary conservation is playing an increasingly important role in maintaining ecosystem integrity and halting biodiversity loss caused by anthropogenic activities. However, lack of information on species distributions in transboundary regions and understanding of the threats in these areas impairs conservation. We developed a spatial conservation plan for the transboundary areas between Yunnan province, southwestern China, and neighboring Myanmar, Laos, and Vietnam in the Indo-Burma biodiversity hotspot. To identify priority areas for conservation and restoration, we determined species distribution patterns and recent land-use changes and examined the spatiotemporal dynamics of the connected natural forest, which supports most species. We assessed connectivity with equivalent connected area (ECA), which is the amount of reachable habitat for a species. An ECA incorporates the presence of habitat in a patch and the amount of habitat in other patches within dispersal distance. We analyzed 197,845 locality records from specimen collections and monographs for 21,004 plant and vertebrate species. The region of Yunnan immediately adjacent to the international borders had the highest species richness, with 61% of recorded species and 56% of threatened vertebrates, which suggests high conservation value. Satellite imagery showed the area of natural forest in the border zone declined by 5.2% (13,255 km2 ) from 1995 to 2018 and monoculture plantations increased 92.4%, shrubland 10.1%, and other cropland 6.2%. The resulting decline in connected natural forest reduced the amount of habitat, especially for forest specialists with limited dispersal abilities. The most severe decline in connectivity was along the Sino-Vietnamese border. Many priority areas straddle international boundaries, indicating demand and potential for establishing transboundary protected areas. Our results illustrate the importance of bi- and multilateral cooperation to protect biodiversity in this region and provide guidance for future conservation planning and practice.

Planeación de la Conservación en las Fronteras de China con Myanmar, Laos y Vietnam Resumen La conservación transfronteriza cada vez juega un papel más importante en la preservación de la integridad del ecosistema y en el freno a la pérdida local de la biodiversidad causada por las actividades antropogénicas. Sin embargo, la falta de información sobre la distribución de las especies en las regiones transfronterizas y de la comprensión de las amenazas en estas áreas obstaculiza la conservación. Desarrollamos un plan de conservación espacial para las áreas transfronterizas entre la provincia de Yunnan, al suroeste de China, y los países vecinos Myanmar, Laos y Vietnam localizadas en el punto caliente de biodiversidad Indo-Burma. Para identificar las áreas prioritarias para la conservación y la restauración, determinamos los patrones de distribución de las especies y los recientes cambios en el uso de suelo y examinamos las dinámicas espaciotemporales del bosque natural conectado, el cual mantiene a la mayoría de las especies. Evaluamos la conectividad con el área equivalente conectada (AEC), que es la cantidad de hábitat accesible para una especie. Un AEC incorpora la presencia del hábitat en un fragmento y la cantidad de hábitat en otros fragmentos dentro de la distancia de dispersión. Analizamos 197,845 registros de localidades desde colecciones de especímenes y monografías para 21,004 especies de plantas y de vertebrados. La región de Yunnan inmediatamente adyacente a las fronteras internacionales tuvo la riqueza de especies más alta con el 61% de las especies registradas y el 56% de los vertebrados amenazados, lo que sugiere un elevado valor de conservación. Las imágenes satelitales mostraron que el área del bosque natural en la zona fronteriza declinó en un 5.2% (13,255 km2 ) entre 1995 y 2018 y que los sembradíos de monocultivos incrementaron en un 92.4%, los matorrales en un 10.1% y otras tierras de cultivo en un 6.2%. La declinación resultante en el bosque natural conectado redujo la cantidad del hábitat, especialmente para los especialistas del bosque con habilidades limitadas de dispersión. La declinación más grave en la conectividad ocurrió a lo largo de la frontera entre China y Vietnam. Muchas áreas prioritarias atraviesan las fronteras internacionales, lo que indica una demanda y un potencial para el establecimiento de áreas protegidas transfronterizas. Nuestros resultados ejemplifican la importancia de la cooperación bi- y multilateral para proteger la biodiversidad en esta región y proporciona información para la planeación y práctica de la conservación en el futuro.

Differences between the perspectives of physicians and patients on the potential barriers to optimal diabetes control in China: a multicenter study.

BACKGROUND: To investigate the potential barriers to optimal diabetes control by evaluating the different perspectives of physicians and patients on such matters in China. METHODS: This multi-center survey was conducted from December 2015 to March 2016. A multi-stage stratified random sampling method was used to sample representative diabetes physicians and patients in 18 hospitals in Shaanxi province, China. A self-designed questionnaire was used. The questionnaire mainly consisted of 2 questions for physicians and 1 question for patients of which the participants were required to rank in priority of 3 (for physicians) and 2 (for patients) choices from a list of barriers. The strategies to improve diabetes control were only in the questionnaire for physicians. RESULTS: A total of 85 physicians and 584 patients completed the questionnaire. Physicians and patients differed regarding the patients' awareness of the risk of diabetes: over 70% of the physicians believed that the patients had no sufficient understanding of the harm and risk of diabetes, whereas the patients believed otherwise. Both physicians and patients considered self-monitoring of blood glucose to be an important link of glucose control; unfortunately, most of the patients failed to do so in practice. In addition, physicians considered "improving health insurance coverage for diabetes" as the first important measure and "providing more and easy-to-use diabetes brochures or educational materials for patients" as the second important measure to improve diabetes control. CONCLUSION: The survey revealed differences between the perspectives of physicians and patients on the potential barriers to optimal diabetes control. The main potential barriers to optimal diabetes control were patient's poor lifestyle interventions, limited understanding of the danger of diabetes, and poor self-monitoring of blood glucose. From the physicians' perspective, China's primary focus about diabetes control in the future should still be put on diabetes education, particular the importance of lifestyle interventions.

Efficacy and safety of endoscopic submucosal tunnel dissection for superficial esophageal squamous cell carcinoma: a propensity score matching analysis.

BACKGROUND AND AIMS: Esophageal endoscopic submucosal dissection (ESD) is technically challenging because of the thinner wall and narrow lumen. The tunnel technique was proposed previously. This current retrospective study aimed to evaluate the efficacy of the tunnel technique in ESD of superficial esophageal squamous cell carcinoma (ESCC). METHODS: Patients who underwent ESD for superficial ESCC between October 2013 and September 2015 were included in the study. Propensity score matching was used to compensate for the differences in age, sex, resected specimen size, and pathology. Treatment outcomes were compared with conventional statistic methods between the tunnel ESD group and conventional ESD group after matching. To further explore the potential variables relevant to procedure time, univariate and multivariate logistic regression analyses were applied. RESULTS: A total of 115 lesions were included in the analysis. Propensity score matching analysis created 38 matched pairs. There was no difference on en bloc resection rate, complete resection rate, and curative rate between the 2 groups. The ESD procedure time was 38.0 (range 29.5-46.0) minutes in the tunnel ESD group and 48.0 (35.4-83.3) minutes in the conventional ESD group (P = .006). There was no difference in adverse events including postprocedural bleeding, perforation, and chest pain, but a lower rate of muscular injury (28.9% vs 52.6%; P = .036) and a less-frequent use of coagulation forceps (36.8% vs 65.8%; P = .012) were shown in the tunnel ESD group. In multivariate regression analysis for procedure time, the tunnel ESD technique (odds ratio [OR] 3.42; 95% confidence interval [CI], 1.32-8.85; P = .011) and specimen size <40 mm (OR 8.74; 95% CI, 1.30-58.5; P = .026) were associated with a shorter procedure time. CONCLUSIONS: The endoscopic submucosal tunnel dissection improved the efficacy and safety of the ESD procedure by shortening the procedure time and reducing injury to the muscular layer.

[Study on analgesia of oxymatrine and its relation to calcium channels].

OBJECTIVE: To study whether the analgesis of oxymatrine (OMT) affects N-type voltage-gated calcium channels (VGCCs). METHODS: Totally 45 mice were randomly divided into the sham-operation group, the model group [established by partial sciatic nerve ligation (PSNL)] , and the OMT treatment group according to random digit table, 15 in each group. The dorsal root ganglions (DRG) were separated in PSNL pain model mice. Intracellular calcium concentration ([Ca2+]i) was determined with Fluo-3 AM immunofluorescent probe in cultured DRG neurons. Different protein expression levels of N-type (Cav2. 2) and L-type ( Cav1. 3) among VGCCs from brain and DRG tissues were detected with Western blot. RESULTS: Compared with the sham-operation group, [Ca2+]i, increased in cultured DRG neurons (P <0. 05) , protein expression levels of Cav2. 2 in the brain tissue increased (P <0. 05), protein expression levels of Cav2. 2 in DRG tissues decreased in the model group (P <0. 01). Compared with the model group, [Ca2+]i, decreased in cultured DRG neurons (P < 0. 05), protein expression levels of Cav2. 2 in the brain tissue decreased (P <0. 01), protein expression levels of Cav2. 2 in DRG tissues increased in the OMT treatment group (P <0. 01). There was no statistical difference in Cav1. 3 expressions in cultured DRG neurons and the brain (P >0. 05). CONCLUSION: Analgesic effect of OMT might be related to Cav2. 2 channel mediated calcium ion flux.

[The Analgesia of Oxymatrine Affecting Calcium Channel and GABA Release].

OBJECTIVE: To explore the analgesia of oxymatrine (OMT) affecting high voltage-dependent calcium channels (HVDCCs) and GABA release under neuropathic pain condition. METHODS: Totally 66 C57BL/6 mice were randomly divided into the sham-operation group, the model group, and the OMT group, 22 in each group. Neuropathic pain models were established by partial sciatic nerve ligation (PSNL). Hind paw plantar mechanical response threshold (MWT) was measured by up-and-down method with Von-Frey filament. mRNA expression of HVDCCs in brains and spinal cords was detected with Real-time PCR and concentration of GABA was determined using ELISA kit. RESULTS: Compared with day 0, the left hind paw MWTwas decreased on day 7, 10, and 14 in the model group (P < 0.05). Compared with the sham-operation group, the left hind paw MWT was significantly reduced in the model group on day 7 (P < 0.05). The MWT of PSNL ipsilateral hind paw was decreased on day 7 before OMT administration, when compared with day 0 (P < 0.05), and increased after OMT administration (P < 0.05). Compared with the sham-operation group, mRNA levels of Cav1.2, Cav1.3, Cav2.1, and Cav2.3 in brain tissues were increased and those of Cav2.2 were decreased significantly in the model group (P < 0.05). In spinal cord tissues, mRNA levels of Cav1.2 and Cav1.3 were increased, but those of Cav2.1, Cav2.2, and Cav2. 3 were decreased significantly in the model group, when compared with those of the sham-operation group (P < 0.05). Compared with the model group, mRNA levels of Cavl.2, Cavl.3, Cav2.1, and Cav2. 3 in brain tissues were decreased, and those of Cav2.2 were increased significantly in the OMT group (P < 0.05). In spinal cord tissues of the OMT group, mRNA levels of Cav1.3 decreased and those of Cav2.1, Cav2.2, and Cav2.3 increased significantly with statistical difference, when compared with those of the model group (P < 0.05). Compared with the sham-operation group, GABA levels in brain tissues decreased in the model group (P < 0.05). Compared with the model group, GABA levels in brain tissues increased in the OMT group (P < 0.05). There was no statistical difference in GABA levels of spinal cord tissues among these groups (P > 0.05). CONCLUSIONS: OMT had analgesic effect on neuropathic pain, which might be probably related to HVDDCs. Cav2.2 might directly affect GABA release.

Optimizing tradeoff strategies of soil microbial community between metabolic efficiency and resource acquisition along a natural regeneration chronosequence.

The tradeoff between community-level soil microbial metabolic efficiency and resource acquisition strategies during natural regeneration remains unclear. Herein, we examined variations in soil extracellular enzyme activity, microbial metabolic quotient (qCO2), and microbial carbon use efficiency (CUE) along a chronosequence of natural regeneration by sampling secondary forests at 1, 10, 20, 30, 40, and 100 years after rubber plantation (RP) clearance. The results showed that the natural logarithms of carbon (C)-, nitrogen (N)-, and phosphorus (P)-acquiring enzyme activities were 1:1.68:1.37 and 1:1.54:1.38 in the RP and secondary forests, respectively, thus demonstrating that microbial metabolism was co-limited by N and P. Moreover, the soil microbial C limitation initially increased (1-40 years) and later decreased (100 years). Overall, the qCO2 increased, decreased, and then increased again in the initial (< 10 years), middle (10-40 years), and late (100 years) successional stages, respectively. Except for specific P-acquiring enzyme activities, the changes in other indicators with natural regeneration were consistent in the dry and wet seasons. Both qCO2 and CUE were mainly predicted by microbial community composition and physiological traits. These results indicate that soil microbial communities could employ tradeoff strategies between metabolic efficiency and resource acquisition to cope with variations in resources. Our findings provide new information on tradeoff strategies between metabolic efficiency and resource acquisition during natural regeneration.

Insecticide Susceptibility and KDR Mutations in Aedes Albopictus Collected from Seven Districts of Guangyuan city, Northern Sichuan, China.

The Asian tiger mosquito, Aedes albopictus, is an important vector of chikungunya, dengue, yellow fever, and Zika viruses. Vector control remains an important means for the prevention and control of vector-borne diseases. The development of insecticide resistance has become a serious threat to the efficacy of insecticide-based control programs. To understand the resistance status and the underlying genetic mechanism in mosquitoes in Guangyuan City of Sichuan Province, China, we investigated the susceptibility of Ae. albopictus to four commonly used insecticides. We found that all the examined populations were susceptible to malathion and propoxur. However, Ae. albopictus populations in Guangyuan showed a possible resistance to the two tested pyrethroids (beta-cypermethrin and deltamethrin). Notably, phenotypic resistance to deltamethrin was detected in 2 of the 7 populations. The potential of resistance to pyrethroids was confirmed by the presence of knockdown resistance (kdr) related mutations in the voltage-gated sodium channel. Four kdr mutations (V1016G, I1532T, F1534L, and F1534S) were identified to be present alone or in combination, and their distribution displayed significant spatial heterogeneity. These findings are helpful for making evidence-based mosquito control strategies and highlight the need to regularly monitor the dynamics of pyrethroid resistance in this city.

Rotation-Invariant Attention Network for Hyperspectral Image Classification.

Hyperspectral image (HSI) classification refers to identifying land-cover categories of pixels based on spectral signatures and spatial information of HSIs. In recent deep learning-based methods, to explore the spatial information of HSIs, the HSI patch is usually cropped from original HSI as the input. And 3 ×3 convolution is utilized as a key component to capture spatial features for HSI classification. However, the 3 ×3 convolution is sensitive to the spatial rotation of inputs, which results in that recent methods perform worse in rotated HSIs. To alleviate this problem, a rotation-invariant attention network (RIAN) is proposed for HSI classification. First, a center spectral attention (CSpeA) module is designed to avoid the influence of other categories of pixels to suppress redundant spectral bands. Then, a rectified spatial attention (RSpaA) module is proposed to replace 3 ×3 convolution for extracting rotation-invariant spectral-spatial features from HSI patches. The CSpeA module, the 1 ×1 convolution and the RSpaA module are utilized to build the proposed RIAN for HSI classification. Experimental results demonstrate that RIAN is invariant to the spatial rotation of HSIs and has superior performance, e.g., achieving an overall accuracy of 86.53% (1.04% improvement) on the Houston database. The codes of this work are available at https://github.com/spectralpublic/RIAN.

Reconstructive Sequence-Graph Network for Video Summarization.

Exploiting the inner-shot and inter-shot dependencies is essential for key-shot based video summarization. Current approaches mainly devote to modeling the video as a frame sequence by recurrent neural networks. However, one potential limitation of the sequence models is that they focus on capturing local neighborhood dependencies while the high-order dependencies in long distance are not fully exploited. In general, the frames in each shot record a certain activity and vary smoothly over time, but the multi-hop relationships occur frequently among shots. In this case, both the local and global dependencies are important for understanding the video content. Motivated by this point, we propose a reconstructive sequence-graph network (RSGN) to encode the frames and shots as sequence and graph hierarchically, where the frame-level dependencies are encoded by long short-term memory (LSTM), and the shot-level dependencies are captured by the graph convolutional network (GCN). Then, the videos are summarized by exploiting both the local and global dependencies among shots. Besides, a reconstructor is developed to reward the summary generator, so that the generator can be optimized in an unsupervised manner, which can avert the lack of annotated data in video summarization. Furthermore, under the guidance of reconstruction loss, the predicted summary can better preserve the main video content and shot-level dependencies. Practically, the experimental results on three popular datasets (i.e., SumMe, TVsum and VTW) have demonstrated the superiority of our proposed approach to the summarization task.

Visible-Infrared Person Re-Identification via Partially Interactive Collaboration.

Visible-infrared person re-identification (VI-ReID) task aims to retrieve the same person between visible and infrared images. VI-ReID is challenging as the images captured by different spectra present large cross-modality discrepancy. Many methods adopt a two-stream network and design additional constraint conditions to extract shared features for different modalities. However, the interaction between the feature extraction processes of different modalities is rarely considered. In this paper, a partially interactive collaboration method is proposed to exploit the complementary information of different modalities to reduce the modality gap for VI-ReID. Specifically, the proposed method is achieved in a partially interactive-shared architecture: collaborative shallow layers and shared deep layers. The collaborative shallow layers consider the interaction between modality-specific features of different modalities, encouraging the feature extraction processes of different modalities constrain each other to enhance feature representations. The shared deep layers further embed the modality-specific features to a common space to endow them the same identity discriminability. To ensure the interactive collaborative learning implement effectively, the conventional loss and collaborative loss are utilized jointly to train the whole network. Extensive experiments on two publicly available VI-ReID datasets verify the superiority of the proposed PIC method. Specifically, the proposed method achieves a rank-1 accuracy of 83.6% and 57.5% on RegDB and SYSU-MM01 datasets, respectively.

Semisupervised Consistent Projection Metric Learning for Person Reidentification.

Person reidentification is a hot topic in the computer vision field. Many efforts have been paid on modeling a discriminative distance metric. However, existing metric-learning-based methods are a lack of generalization. In this article, the poor generalization of the metric model is argued as the biased estimation problem that the independent identical distribution hypothesis is not valid. The verification experimental result shows that there is a sharp difference between the training and test samples in the metric subspace. A semisupervised consistent projection metric-learning method is proposed to ease the biased estimation problem by learning a consistent constrained metric subspace in which the identified pairs are forced to follow the distribution of the positive training pairs. First, a semisupervised method is proposed to generate potential matching pairs from the k -nearest neighbors of test samples. The potential matching pairs are used to estimate the distances' distribution center of the positive test pairs. Second, the metric subspace is improved by forcing this estimation to be close to the center of the positive training pairs. Finally, extensive experiments are conducted on five datasets and the results demonstrate that the proposed method reaches the best performance, especially on the rank-1 identification rate.

Deep Category-Level and Regularized Hashing With Global Semantic Similarity Learning.

The hashing technique has been extensively used in large-scale image retrieval applications due to its low storage and fast computing speed. Most existing deep hashing approaches cannot fully consider the global semantic similarity and category-level semantic information, which result in the insufficient utilization of the global semantic similarity for hash codes learning and the semantic information loss of hash codes. To tackle these issues, we propose a novel deep hashing approach with triplet labels, namely, deep category-level and regularized hashing (DCRH), to leverage the global semantic similarity of deep feature and category-level semantic information to enhance the semantic similarity of hash codes. There are four contributions in this article. First, we design a novel global semantic similarity constraint about the deep feature to make the anchor deep feature more similar to the positive deep feature than to the negative deep feature. Second, we leverage label information to enhance category-level semantics of hash codes for hash codes learning. Third, we develop a new triplet construction module to select good image triplets for effective hash functions learning. Finally, we propose a new triplet regularized loss (Reg-L) term, which can force binary-like codes to approximate binary codes and eventually minimize the information loss between binary-like codes and binary codes. Extensive experimental results in three image retrieval benchmark datasets show that the proposed DCRH approach achieves superior performance over other state-of-the-art hashing approaches.

A Supervised Segmentation Network for Hyperspectral Image Classification.

Recently, deep learning has drawn broad attention in the hyperspectral image (HSI) classification task. Many works have focused on elaborately designing various spectral-spatial networks, where convolutional neural network (CNN) is one of the most popular structures. To explore the spatial information for HSI classification, pixels with its adjacent pixels are usually directly cropped from hyperspectral data to form HSI cubes in CNN-based methods. However, the spatial land-cover distributions of cropped HSI cubes are usually complicated. The land-cover label of a cropped HSI cube cannot simply be determined by its center pixel. In addition, the spatial land-cover distribution of a cropped HSI cube is fixed and has less diversity. For CNN-based methods, training with cropped HSI cubes will result in poor generalization to the changes of spatial land-cover distributions. In this paper, an end-to-end fully convolutional segmentation network (FCSN) is proposed to simultaneously identify land-cover labels of all pixels in a HSI cube. First, several experiments are conducted to demonstrate that recent CNN-based methods show the weak generalization capabilities. Second, a fine label style is proposed to label all pixels of HSI cubes to provide detailed spatial land-cover distributions of HSI cubes. Third, a HSI cube generation method is proposed to generate plentiful HSI cubes with fine labels to improve the diversity of spatial land-cover distributions. Finally, a FCSN is proposed to explore spectral-spatial features from finely labeled HSI cubes for HSI classification. Experimental results show that FCSN has the superior generalization capability to the changes of spatial land-cover distributions.

Bidirectional Interaction Network for Person Re-Identification.

Person re-identification (ReID) task aims to retrieve the same person across multiple spatially disjoint camera views. Due to huge image changes caused by various factors such as posture variation and illumination transformation, images of different persons may share the more similar appearances than images of the same one. Learning discriminative representations to distinguish details of different persons is significant for person ReID. Many existing methods learn discriminative representations resorting to a human body part location branch which requires cumbersome expert human annotations or complex network designs. In this article, a novel bidirectional interaction network is proposed to explore discriminative representations for person ReID without any human body part detection. The proposed method regards multiple convolutional features as responses to various body part properties and exploits the inter-layer interaction to mine discriminative representations for person identities. Firstly, an inter-layer bilinear pooling strategy is proposed to feasibly exploit the pairwise feature relations between two convolution layers. Secondly, to explore interaction of multiple layers, an effective bidirectional integration strategy consisting of two different multi-layer interaction processes is designed to aggregate bilinear pooling interaction of multiple convolution layers. The interaction of multiple layers is implemented in a layer-by-layer nesting policy to ensure the two interaction processes are different and complementary. Extensive experiments validate the superiority of the proposed method on four popular person ReID datasets including Market-1501, DukeMTMC-ReID, CUHK03-NP and MSMT17. Specifically, the proposed method achieves a rank-1 accuracy of 95.1% and 88.2% on Market-1501 and DukeMTMC-ReID, respectively.

Person Reidentification via Unsupervised Cross-View Metric Learning.

Person reidentification (Re-ID) aims to match observations of individuals across multiple nonoverlapping camera views. Recently, metric learning-based methods have played important roles in addressing this task. However, metrics are mostly learned in supervised manners, of which the performance relies heavily on the quantity and quality of manual annotations. Meanwhile, metric learning-based algorithms generally project person features into a common subspace, in which the extracted features are shared by all views. However, it may result in information loss since these algorithms neglect the view-specific features. Besides, they assume person samples of different views are taken from the same distribution. Conversely, these samples are more likely to obey different distributions due to view condition changes. To this end, this paper proposes an unsupervised cross-view metric learning method based on the properties of data distributions. Specifically, person samples in each view are taken from a mixture of two distributions: one models common prosperities among camera views and the other focuses on view-specific properties. Based on this, we introduce a shared mapping to explore the shared features. Meanwhile, we construct view-specific mappings to extract and project view-related features into a common subspace. As a result, samples in the transformed subspace follow the same distribution and are equipped with comprehensive representations. In this paper, these mappings are learned in an unsupervised manner by clustering samples in the projected space. Experimental results on five cross-view datasets validate the effectiveness of the proposed method.

Bio-Inspired Representation Learning for Visual Attention Prediction.

Visual attention prediction (VAP) is a significant and imperative issue in the field of computer vision. Most of the existing VAP methods are based on deep learning. However, they do not fully take advantage of the low-level contrast features while generating the visual attention map. In this article, a novel VAP method is proposed to generate the visual attention map via bio-inspired representation learning. The bio-inspired representation learning combines both low-level contrast and high-level semantic features simultaneously, which are developed by the fact that the human eye is sensitive to the patches with high contrast and objects with high semantics. The proposed method is composed of three main steps: 1) feature extraction; 2) bio-inspired representation learning; and 3) visual attention map generation. First, the high-level semantic feature is extracted from the refined VGG16, while the low-level contrast feature is extracted by the proposed contrast feature extraction block in a deep network. Second, during bio-inspired representation learning, both the extracted low-level contrast and high-level semantic features are combined by the designed densely connected block, which is proposed to concatenate various features scale by scale. Finally, the weighted-fusion layer is exploited to generate the ultimate visual attention map based on the obtained representations after bio-inspired representation learning. Extensive experiments are performed to demonstrate the effectiveness of the proposed method.

Vision-to-Language Tasks Based on Attributes and Attention Mechanism.

Vision-to-language tasks aim to integrate computer vision and natural language processing together, which has attracted the attention of many researchers. For typical approaches, they encode image into feature representations and decode it into natural language sentences. While they neglect high-level semantic concepts and subtle relationships between image regions and natural language elements. To make full use of these information, this paper attempt to exploit the text-guided attention and semantic-guided attention (SA) to find the more correlated spatial information and reduce the semantic gap between vision and language. Our method includes two-level attention networks. One is the text-guided attention network which is used to select the text-related regions. The other is SA network which is used to highlight the concept-related regions and the region-related concepts. At last, all these information are incorporated to generate captions or answers. Practically, image captioning and visual question answering experiments have been carried out, and the experimental results have shown the excellent performance of the proposed approach.

Revision of Nagiella Munroe (Lepidoptera, Crambidae), with the description of a new species from China.

The genus Nagiella was studied using morphological and DNA barcode data. Nagiella bispina sp. nov. is described as a new species, and N. hortulatoides Munroe is recorded in China for the first time. The diagnosis of this genus is revised, and the genitalia description of N. quadrimaculalis (Kollar and Redtenbacher) and N. inferior (Hampson) are given in English for the first time. Nosophora incomitata (Swinhoe) stat. rev. is removed from the synonym of N. quadrimaculalis. Photographs of the habitus and genitalia as well as COI DNA Barcode data of these four species are provided.

A Joint Relationship Aware Neural Network for Single-Image 3D Human Pose Estimation.

This paper studies the task of 3D human pose estimation from a single RGB image, which is challenging without depth information. Recently many deep learning methods are proposed and achieve great improvements due to their strong representation learning. However, most existing methods ignore the relationship between joint features. In this paper, a joint relationship aware neural network is proposed to take both global and local joint relationship into consideration. First, a whole feature block representing all human body joints is extracted by a convolutional neural network. A Dual Attention Module (DAM) is applied on the whole feature block to generate attention weights. By exploiting the attention module, the global relationship between the whole joints is encoded. Second, the weighted whole feature block is divided into some individual joint features. To capture salient joint feature, the individual joint features are refined by individual DAMs. Finally, a joint angle prediction constraint is proposed to consider local joint relationship. Quantitative and qualitative experiments on 3D human pose estimation benchmarks demonstrate the effectiveness of the proposed method.

Discrete Deep Hashing With Ranking Optimization for Image Retrieval.

For large-scale image retrieval task, a hashing technique has attracted extensive attention due to its efficient computing and applying. By using the hashing technique in image retrieval, it is crucial to generate discrete hash codes and preserve the neighborhood ranking information simultaneously. However, both related steps are treated independently in most of the existing deep hashing methods, which lead to the loss of key category-level information in the discretization process and the decrease in discriminative ranking relationship. In order to generate discrete hash codes with notable discriminative information, we integrate the discretization process and the ranking process into one architecture. Motivated by this idea, a novel ranking optimization discrete hashing (RODH) method is proposed, which directly generates discrete hash codes (e.g., +1/-1) from raw images by balancing the effective category-level information of discretization and the discrimination of ranking information. The proposed method integrates convolutional neural network, discrete hash function learning, and ranking function optimizing into a unified framework. Meanwhile, a novel loss function based on label information and mean average precision (MAP) is proposed to preserve the label consistency and optimize the ranking information of hash codes simultaneously. Experimental results on four benchmark data sets demonstrate that RODH can achieve superior performance over the state-of-the-art hashing methods.