Despite significant efforts made so far for Weakly Supervised Object Detection (WSOD), proposal generation and proposal selection are still two major challenges. In this paper, we focus on addressing ...the two challenges by generating and selecting high-quality proposals. To be specific, for proposal generation, we combine selective search and a Gradient-weighted Class Activation Mapping (Grad-CAM) based technique to generate more proposals having higher Intersection-Over-Union (IOU) with ground truth boxes than those obtained by greedy search approaches, which can better envelop the entire objects. As regards proposal selection, for each object class, we choose as many confident positive proposals as possible and meanwhile only select class-specific hard negatives to focus training on more discriminative negative proposals by up-weighting their losses, which can make training more effective. The proposed proposal generation and proposal selection approaches are generic and thus can be broadly applied to many WSOD methods. In this work, we unify them into the framework of Online Instance Classifier Refinement (OICR). Experimental results on the PASCAL VOC 2007 and 2012 datasets and MS COCO dataset demonstrate that our method significantly improves the baseline method OICR by large margins (13.4% mAP and 11.6% CorLoc gains on the VOC 2007 dataset, 15.0% mAP and 8.9% CorLoc gains on the VOC 2012 dataset, and 6.4% mAP and 5.0% CorLoc gains on the COCO dataset) and achieves the state-of-the-art results compared with existing methods.
Abstract O-GlcNAcylation is the posttranslational modification of intracellular proteins by O-linked β-N-acetylglucosamine (O-GlcNAc). The discovery of O-GlcNAc modification of tau and its impact on ...tau phosphorylation has attracted recent research interest in O-GlcNAc studies in the Alzheimer's disease (AD) field. Modification of proteins by O-GlcNAc occurs extensively in the brain. The expressions and activities of the enzymes catalyzing O-GlcNAc cycling are several-fold higher in the brain than in the peripheral tissues. The O-GlcNAcylation levels of brain proteins including tau are decreased in AD brain, probably due to decreased brain glucose metabolism. The reduction of brain O-GlcNAcylation appears to mediate the molecular mechanism by which decreased brain glucose metabolism contributes to neurodegeneration. Studies on mouse models of tauopathies suggest a neuroprotective role of pharmacological elevation of brain O-GlcNAc, which could potentially be a promising approach for treating AD and other neurodegenerative diseases.
Various studies have shown that soils surrounding mining areas are seriously polluted with heavy metals. Determining the effects of natural factors on spatial distribution of heavy metals is ...important for determining the distribution characteristics of heavy metals in soils. In this study, an 8km buffer zone surrounding a typical non-ferrous metal mine in Suxian District of Hunan Province, China, was selected as the study area, and statistical, spatial autocorrelation and spatial interpolation analyses were used to obtain descriptive statistics and spatial autocorrelation characteristics of As, Pb, Cu, and Zn in soil. Additionally, the distributions of soil heavy metals under the influences of natural factors, including terrain (elevation and slope), wind direction and distance from a river, were determined. Layout of sampling sites, spatial changes of heavy metal contents at high elevations and concentration differences between upwind and downwind directions were then evaluated. The following results were obtained: (1) At low elevations, heavy metal concentrations decreased slightly, then increased considerably with increasing elevation. At high elevations, heavy metal concentrations first decreased, then increased, then decreased with increasing elevation. As the slope increased, heavy metal contents increased then decreased. (2) Heavy metal contents changed consistently in the upwind and downwind directions. Heavy metal contents were highest in 1km buffer zone and decreased with increasing distance from the mining area. The largest decrease in heavy metal concentrations was in 2km buffer zone. Perennial wind promotes the transport of heavy metals in downwind direction. (3) The spatial extent of the influence of the river on Pb, Zn and Cu in the soil was 800m. (4) The influence of the terrain on the heavy metal concentrations was greater than that of the wind. These results provide a scientific basis for preventing and mitigating heavy metal soil pollution in areas surrounding mines.
Display omitted
•Distributions of As, Pb, Cu, Zn under influences of natural factors are explored.•Distributions of heavy metals with slopes are different from that of elevations.•Perennial wind promotes the spread of heavy metal to downwind direction.•Influence ranges of river on Pb, Zn and Cu in surrounding soil are within 800m.•Influence of terrain on distributions of heavy metals is greater than wind.
In the past few years, object detection in remote sensing images has achieved remarkable progress. However, the detection of oriented and densely packed objects are still unsatisfactory due to the ...following spatial and feature misalignments. 1) Most two-stage oriented detectors only introduce an orientation regression branch in the detection head, while still leverage horizontal proposals for classification and regression. This inevitably results in the spatial misalignment problem between horizontal proposals and oriented objects. 2) The features used for classification are in fact extracted from the region proposals which have shifted to the final predictions via the regression branch. This leads to the feature misalignment problem between the classification and the localization tasks. In this article, we present a two-stage oriented object detection method, termed dual-aligned oriented detector (DODet), toward evading the aforementioned problems of spatial and feature misalignments. In DODet, the first stage is an oriented proposal network (OPN), which generates high-quality oriented proposals via a novel representation scheme of oriented objects. The second stage is a localization-guided detection head (LDH) that aims at alleviating the feature misalignment between classification and localization. Comprehensive and extensive evaluations on three benchmarks, including DIOR-R, DOTA, and HRSC2016, indicate that our method could obtain consistent and substantial gains compared with the baseline method. The source code is publicly available at https://github.com/yanqingyao1994/DODet .
Display omitted
•Bio-inspired solar steam generation using floating plasmonic membranes (PMs) proposed.•High steam generation efficiency of 85% achieved at illumination power of 10kWm−2.•PMs enhanced ...the productivity of a solar still for seawater desalination by ∼80%.
Efficient solar-enabled evaporation plays a critical role in solar power-based concentration systems, photochemical plants, seawater desalination technologies, etc. However, traditional processes for solar steam generation usually depend on high-temperature heating of the bulk liquid, which requires highly concentrated solar power and suffers from high energy and optical losses. Therefore, the enhancement of solar steam generation by bio-inspired interface solar heating is proposed in this work. In this study, easy-to-prepare, flexible, and reusable plasmonic membranes (PMs) were fabricated for realizing the bio-inspired interface solar heating and continuous steam transportation through the micropores of the membranes. A solar steam generation efficiency of ∼85% was achieved at an illumination power of 10kWm−2. The effects of Au concentration in the membranes and optical power on the steam generation efficiency were systemically studied. The observed high evaporation rate and efficiency were attributed to three main factors: high (∼90%) and broadband solar absorption, efficient photo-thermal conversion due to high plasmon dissipation losses, and fast capillary flow in the membrane micropores. Finally, the application of PMs in a single basin solar still system for seawater desalination was investigated and the PMs exhibited great performance on enhancing the productivity of clean water.
The objective of detection in remote sensing images is to determine the location and category of all targets in these images. The anchor based methods are the most prevalent deep learning based ...methods, and still have some problems that need to be addressed. First, the existing metric (i.e., intersection over union (IoU)) could not measure the distance between two bounding boxes when they are nonoverlapping. Second, the exsiting bounding box regression loss could not directly optimize the metric in the training process. Third, the existing methods which adopt a hierarchical deep network only choose a single level feature layer for the feature extraction of region proposals, meaning they do not take full use of the advantage of multi-level features. To resolve the above problems, a novel object detection method for remote sensing images based on improved bounding box regression and multi-level features fusion is proposed in this paper. First, a new metric named generalized IoU is applied, which can quantify the distance between two bounding boxes, regardless of whether they are overlapping or not. Second, a novel bounding box regression loss is proposed, which can not only optimize the new metric (i.e., generalized IoU) directly but also overcome the problem that existing bounding box regression loss based on the new metric cannot adaptively change the gradient based on the metric value. Finally, a multi-level features fusion module is proposed and incorporated into the existing hierarchical deep network, which can make full use of the multi-level features for each region proposal. The quantitative comparisons between the proposed method and baseline method on the large scale dataset DIOR demonstrate that incorporating the proposed bounding box regression loss, multi-level features fusion module, and a combination of both into the baseline method can obtain an absolute gain of 0.7%, 1.4%, and 2.2% or so in terms of mAP, respectively. Comparing this with the state-of-the-art methods demonstrates that the proposed method has achieved a state-of-the-art performance. The curves of average precision with different thresholds show that the advantage of the proposed method is more evident when the threshold of generalized IoU (or IoU) is relatively high, which means that the proposed method can improve the precision of object localization. Similar conclusions can be obtained on a NWPU VHR-10 dataset.
Duplex Metric Learning for Image Set Classification Cheng, Gong; Zhou, Peicheng; Han, Junwei
IEEE transactions on image processing,
2018-Jan., 2018-Jan, 2018-1-00, 20180101, Volume:
27, Issue:
1
Journal Article
Peer reviewed
Image set classification has attracted much attention because of its broad applications. Despite the success made so far, the problems of intra-class diversity and inter-class similarity still remain ...two major challenges. To explore a possible solution to these challenges, this paper proposes a novel approach, termed duplex metric learning (DML), for image set classification. The proposed DML consists of two progressive metric learning stages with different objectives used for feature learning and image classification, respectively. The metric learning regularization is not only used to learn powerful feature representations but also well explored to train an effective classifier. At the first stage, we first train a discriminative stacked autoencoder (DSAE) by layer-wisely imposing a metric learning regularization term on the neurons in the hidden layers and meanwhile minimizing the reconstruction error to obtain new feature mappings in which similar samples are mapped closely to each other and dissimilar samples are mapped farther apart. At the second stage, we discriminatively train a classifier and simultaneously fine-tune the DSAE by optimizing a new objective function, which consists of a classification error term and a metric learning regularization term. Finally, two simple voting strategies are devised for image set classification based on the learnt classifier. In the experiments, we extensively evaluate the proposed framework for the tasks of face recognition, object recognition, and face verification on several commonly-used data sets and state-of-the-art results are achieved in comparison with existing methods.
Abstract Abnormal sphingolipid metabolism has been previously reported in Alzheimer's disease (AD). To extend these findings, several sphingolipids and sphingolipid hydrolases were analyzed in brain ...samples from AD patients and age-matched normal individuals. We found a pattern of elevated acid sphingomyelinase (ASM) and acid ceramidase (AC) expression in AD, leading to a reduction in sphingomyelin and elevation of ceramide. More sphingosine also was found in the AD brains, although sphingosine-1-phosphate (S1P) levels were reduced. Notably, significant correlations were observed between the brain ASM and S1P levels and the levels of amyloid beta (Aβ) peptide and hyperphosphorylated tau protein. Based on these findings, neuronal cell cultures were treated with Aβ oligomers, which were found to activate ASM, increase ceramide, and induce apoptosis. Pre-treatment of the neurons with purified, recombinant AC prevented the cells from undergoing Aβ-induced apoptosis. We propose that ASM activation is an important pathological event leading to AD, perhaps due to Aβ deposition. The downstream consequences of ASM activation are elevated ceramide, activation of ceramidases, and production of sphingosine. The reduced levels of S1P in the AD brain, together with elevated ceramide, likely contribute to the disease pathogenesis.
With the transition of the mobile communication networks, the network goal of the Internet of everything further promotes the development of the Internet of Things (IoT) and Wireless Sensor Networks ...(WSNs). Since the directional sensor has the performance advantage of long-term regional monitoring, how to realize coverage optimization of Directional Sensor Networks (DSNs) becomes more important. The coverage optimization of DSNs is usually solved for one of the variables such as sensor azimuth, sensing radius, and time schedule. To reduce the computational complexity, we propose an optimization coverage scheme with a boundary constraint of eliminating redundancy for DSNs. Combined with Particle Swarm Optimization (PSO) algorithm, a Virtual Angle Boundary-aware Particle Swarm Optimization (VAB-PSO) is designed to reduce the computational burden of optimization problems effectively. The VAB-PSO algorithm generates the boundary constraint position between the sensors according to the relationship among the angles of different sensors, thus obtaining the boundary of particle search and restricting the search space of the algorithm. Meanwhile, different particles search in complementary space to improve the overall efficiency. Experimental results show that the proposed algorithm with a boundary constraint can effectively improve the coverage and convergence speed of the algorithm.