Directional Deep Embedding and Appearance Learning for Fast Video Object Segmentation.

Most recent semisupervised video object segmentation (VOS) methods rely on fine-tuning deep convolutional neural networks online using the given mask of the first frame or predicted masks of subsequent frames. However, the online fine-tuning process is usually time-consuming, limiting the practical use of such methods. We propose a directional deep embedding and appearance learning (DDEAL) method, which is free of the online fine-tuning process, for fast VOS. First, a global directional matching module (GDMM), which can be efficiently implemented by parallel convolutional operations, is proposed to learn a semantic pixel-wise embedding as an internal guidance. Second, an effective directional appearance model-based statistics is proposed to represent the target and background on a spherical embedding space for VOS. Equipped with the GDMM and the directional appearance model learning module, DDEAL learns static cues from the labeled first frame and dynamically updates cues of the subsequent frames for object segmentation. Our method exhibits the state-of-the-art VOS performance without using online fine-tuning. Specifically, it achieves a J & F mean score of 74.8% on DAVIS 2017 data set and an overall score G of 71.3% on the large-scale YouTube-VOS data set, while retaining a speed of 25 fps with a single NVIDIA TITAN Xp GPU. Furthermore, our faster version runs 31 fps with only a little accuracy loss.

Goethite modified biochar simultaneously mitigates the arsenic and cadmium accumulation in paddy rice (Oryza sativa) L.

Cadmium (Cd) and arsenic (As) contamination of paddy soils is a serious global issue because of the opposite geochemical behavior of Cd and As in paddy soils. Rice plant (Oryza sativa L.) cultivation in Cd- and As- contaminated paddy soil is regarded as one of the main dietary cause of Cd and As entry in human beings. This study aimed to determine the impact of goethite-modified biochar (GB) on bioavailability of both Cd and As in Cd- and As- polluted paddy soil. Contrary to control and biochar (BC) amendments, the application of GB amendments significantly impeded the accumulation of both Cd and As in rice plants. The results confirmed an obvious reduction in Cd and As content of rice grains by 85% and 77%, respectively after soil supplementation with GB 2% amendment. BC 3% application minimized the Cd uptake by 59% in the rice grains as compared to the control but exhibited a little impact on As accumulation in rice grains. Sequential extraction results displayed an increase in immobile Cd and As fractions of the soil by decreasing the bioavailable fractions of both elements after GB treatments. Fe-plaque formation on the root surfaces was significantly variable (P Ë‚ 0.05) among all the amendments. GB 2% treatment significantly increased the Fe content (10 g kg-1) of root Fe-plaque by 48%, which ultimately enhanced the sequestration of Cd and As by Fe-plaque and minimized the transport of Cd and As in rice plants. Moreover, GB treatments significantly changed the relative abundance of the microbial community in the rice rhizosphere and minimized the metal(loid)s mobility in the soil. The relative abundance of Acidobacteria, Firmicutes and Verrucomicrobia increased with GB 2% treatment while those of Bacteroidetes and Choloroflexi decreased. Our findings confirmed improvement in the rice grains quality regarding enhanced amino acid contents with GB application. Overall, the results of this study demonstrated that GB amendment simultaneously alleviated the Cd and As concentrations in edible parts of rice plant and provided a new valuable method to protect the public health by effectively remediating the co-occurrence of Cd and As in paddy soils.

PAK4 suppresses TNF-induced release of endothelial microparticles in HUVECs cells.

Tumor necrosis factor-α (TNF) is a pro-inflammatory cytokine upregulated in many inflammatory diseases, and a potent inducer of endothelial cell-derived microparticle (EMP) formation. In this study, we identified the protein kinase PAK4 as a key regulator of the TNF-induced EMP release from human umbilical vein endothelial cells (HUVECs). TNF induces dose- and time-dependent EMP release and downregulation of PAK4 and upstream cdc42 in HUVECs. PAK4 suppression or inhibition of its kinase activity increases TNF-induced EMP release and apoptosis in HUVECs, while PAK4 overexpression reduces EMP release and apoptosis in TNF-stimulated cells. Collectively, these data indicate that PAK4 suppresses TNF-induced EMP generation occurring during apoptosis, and suggest that modulation of PAK4 activity may represent a novel approach to suppress the TNF-induced EMP levels in pro-inflammatory disorders and other pathological conditions.

[Application of Goethite Modified Biochar for Arsenic Removal from Aqueous Solution].

To improve the adsorption capacity of wheat biochar (BC) for arsenic (As), wheat stalks were selected as biomass to generate nano-sized goethite modified biochar (Goethite@BC) by co-precipitation. The adsorption capacities of BC, Goethite, and Goethite@BC for As(Ⅲ) were compared. The samples were analyzed by scanning electron microscopy (SEM) along with energy dispersive spectrometry (EDS), Brunauer-Emmett-Teller (BET), Fourier transform infrared (FT-IR), X-ray diffraction (XRD), and X-ray photoelectron spectroscopy (XPS) techniques. The results showed that the nano-goethite coating was uniformly attached to the surface of the BC and improved the surface area and total pore volume of the biochar. The adsorption of As(Ⅲ) by the three adsorbents was proved to fit well with the pseudo-second-order kinetic model and the Langmuir model. Compared to BC, the Goethite@BC increased the adsorption rate of As(Ⅲ) by 62.10 times, and the maximum adsorption capacity of Goethite@BC was 65.20 mg·g-1. The adsorption mechanism of Goethite@BC included non-specific adsorption (electrostatic attraction) and specific adsorption (coordination, complexation, ion exchange, etc.), and nano-goethite particles on the Goethite@BC surface played an important role in the adsorption of As. Goethite@BC has a good application prospects in the field of environmental remediation.

Robust Landmark Detection and Position Measurement Based on Monocular Vision for Autonomous Aerial Refueling of UAVs.

In this paper, a position measurement system, including drogue's landmark detection and position computation for autonomous aerial refueling of unmanned aerial vehicles, is proposed. A multitask parallel deep convolution neural network (MPDCNN) is designed to detect the landmarks of the drogue target. In MPDCNN, two parallel convolution networks are used, and a fusion mechanism is proposed to accomplish the effective fusion of the drogue's two salient parts' landmark detection. Considering the drogue target's geometric constraints, a position measurement method based on monocular vision is proposed. An effective fusion strategy, which fuses the measurement results of drogue's different parts, is proposed to achieve robust position measurement. The error of landmark detection with the proposed method is 3.9%, and it is obviously lower than the errors of other methods. Experimental results on the two KUKA robots platform verify the effectiveness and robustness of the proposed position measurement system for aerial refueling.

Face Detection With Different Scales Based on Faster R-CNN.

In recent years, the application of deep learning based on deep convolutional neural networks has gained great success in face detection. However, one of the remaining open challenges is the detection of small-scaled faces. The depth of the convolutional network can cause the projected feature map for small faces to be quickly shrunk, and most detection approaches with scale invariant can hardly handle less than 15×15 pixel faces. To solve this problem, we propose a different scales face detector (DSFD) based on Faster R-CNN. The new network can improve the precision of face detection while performing as real-time a Faster R-CNN. First, an efficient multitask region proposal network (RPN), combined with boosting face detection, is developed to obtain the human face ROI. Setting the ROI as a constraint, an anchor is inhomogeneously produced on the top feature map by the multitask RPN. A human face proposal is extracted through the anchor combined with facial landmarks. Then, a parallel-type Fast R-CNN network is proposed based on the proposal scale. According to the different percentages they cover on the images, the proposals are assigned to three corresponding Fast R-CNN networks. The three networks are separated through the proposal scales and differ from each other in the weight of feature map concatenation. A variety of strategies is introduced in our face detection network, including multitask learning, feature pyramid, and feature concatenation. Compared to state-of-the-art face detection methods such as UnitBox, HyperFace, FastCNN, the proposed DSFD method achieves promising performance on popular benchmarks including FDDB, AFW, PASCAL faces, and WIDER FACE.

[Changes in circulating microparticles in mice with ventilator-induced lung injury].

OBJECTIVE: To investigate the changes in serum level of microparticles (EMPs) in mice with ventilator-induced lung injury (VILI), and explore its significance in VILI. METHODS: Forty-eight grade SPF male C57BL/6J mice were randomly divided into two groups, with 24 mice in each group: the mice in mechanical ventilation (MV) group were given high tidal volume (VT 30 mL/kg) MV for 4 hours after tracheal intubation, and those in spontaneous breathing group were spontaneously breathed for 4 hours. The apical blood of 12 mice in each group were collected, and serum levels of interleukins (IL-1ß, IL-6) and tumor necrosis factor-α (TNF-α) were determined by enzyme linked immunosorbent assay (ELISA), and serum EMPs levels were determined by flow cytometer. The correlations between EMPs and IL-1ß, IL-6, and TNF-α were analyzed by linear regression analysis. The lung tissues of other 12 mice in each group were harvested, and wet/dry weight (W/D) ratio was assessed. After hematoxylin-eosin (HE) staining, the morphological changes in lung tissue were observed under light microscope. After double staining of uranium acetate and lead citrate, the ultrastructural changes in lung tissue were observed with electron microscope. RESULTS: Compared with spontaneous breathing group, the levels of lung W/D ratio in MV group was significantly increased (5.47±0.14 vs. 4.34±0.11), the levels of IL-1ß, IL-6, TNF-α and EMPs were also significantly increased [IL-1ß (ng/L): 42.4±4.4 vs. 7.7±3.6, IL-6 (ng/L): 1 239.5±66.3 vs. 21.7±4.6, TNF-α (ng/L): 237.6±25.8 vs. 37.1±19.1, EMPs (cells/µL): 28.6±1.8 vs. 5.9±1.8, all P < 0.01]. It was shown by correlation analysis that EMPs were positively related with IL-1ß, IL-6, and TNF-α (r value was 0.968, 0.932, 0.945, respectively, all P = 0.000). It was shown by fitting linear regression analysis that when EMPs increased by 1 cell/µL, IL-1ß increased by 2.4 ng/L [95% confidence interval (95%CI) = 1.9-2.8, P < 0.001], IL-6 increased by 34.5 ng/L (95%CI = 25.1-44.0, P < 0.001), and TNF-α increased by 13.6 ng/L (95%CI = 10.3-16.9,P < 0.001). It was shown by light microscope that the structure of lung tissue and alveolar of mice in spontaneous breathing group appeared normal, while the shrinks of alveolar and disappearance of alveolar architecture were found in MV group. It was shown by electron microscopy that alveolar wall edema and thickening and broken alveolar septa were found in MV group, by contrast, the structure of alveolar was normal in spontaneous breathing group. CONCLUSIONS: 30 mL/kg VT ventilation for 4 hours could induce VILI with increase in EMPs, suggesting EMPs closely related to VILI, and EMPs level may be putative biomarker of VILI.

Online State-Based Structured SVM Combined With Incremental PCA for Robust Visual Tracking.

In this paper, we propose a robust state-based structured support vector machine (SVM) tracking algorithm combined with incremental principal component analysis (PCA). Different from the current structured SVM for tracking, our method directly learns and predicts the object's states and not the 2-D translation transformation during tracking. We define the object's virtual state to combine the state-based structured SVM and incremental PCA. The virtual state is considered as the most confident state of the object in every frame. The incremental PCA is used to update the virtual feature vector corresponding to the virtual state and the principal subspace of the object's feature vectors. In order to improve the accuracy of the prediction, all the feature vectors are projected onto the principal subspace in the learning and prediction process of the state-based structured SVM. Experimental results on several challenging video sequences validate the effectiveness and robustness of our approach.