Search

Object detection is usually solved by learning a deep architecture involving classification and localization tasks, where feature learning for these two tasks is shared using the same backbone model. Recent works have shown that suitable disentanglement of classification and localization tasks has the great potential to improve performance of object detection. Despite the promising performance, existing feature disentanglement methods usually suffer from two limitations. First, most of them only focus on the disentangled proposals or predication heads for classification and localization tasks after RPN. While little consideration has been given to that the features for these two different tasks actually are obtained by a shared backbone model before RPN. Second, they are suggested for two-stage objectors and are not applicable to one-stage methods. To overcome these limitations, this paper presents a novel fully task-specific feature learning method for one-stage object detection. Specifically, our method first learns disentangled features for classification and localization tasks using two separated backbone models, where auxiliary classification and localization heads are inserted at the end of the two backbone models for providing a fully task-specific features for classification and localization. Then, a feature interaction module is developed for aligning and fusing task-specific features, which are further used to produce the final detection result. Experiments on MS COCO show that our proposed method (dubbed CrabNet) can achieve clear improvement over counterparts with increasing limited inference time, while performing favorably against state-of-the-arts. [ABSTRACT FROM AUTHOR]

Weakly supervised semantic segmentation (WSSS) based on bounding box annotations has attracted considerable recent attention and has achieved promising performance. However, most of existing methods focus on generation of high-quality pseudo labels for segmented objects using box indicators, but they fail to fully explore and exploit prior from bounding box annotations, which limits performance of WSSS methods, especially for fine parts and boundaries. To overcome above issues, this paper proposes a novel Pixel-as-Instance Prior (PIP) for WSSS methods by delving deeper into pixel prior from bounding box annotations. Specifically, the proposed PIP is built on two important observations on pixels around bounding boxes. First, since objects are usually irregularity and tightly close to bounding boxes (dubbed irregular-filling prior), so each row or column of bounding boxes basically have at least one pixel belonging to foreground objects and background, respectively. Second, pixels near the bounding boxes tend to be highly ambiguous and more difficult to classify (dubbed label-ambiguity prior). To implement our PIP, a constrained loss alike multiple instance learning (MIL) and a labeling-balance loss are developed to jointly train WSSS models, which regards each pixel as a weighted positive or negative instance while considering more effective prior (i.e., irregular-filling and label-ambiguity priors) from bounding box annotations in an efficient way. Note that our PIP can be flexibly integrated with various WSSS methods, while clearly improving their performance with negligible computational overload in training stage. The experiments are conducted on most widely used PASCAL VOC 2012 and Cityscapes benchmarks, and the results show that our PIP has a good ability to improve performance of various WSSS methods, while achieving very competitive results. [ABSTRACT FROM AUTHOR]

This article explores the potential of epoxy resin/silicon carbide whisker (EP/SiCw) composites (1 wt%, 2 wt%, 3 wt%, 4 wt%, 5 wt%) as the field-dependent conductivity (FDC) packaging materials for the high-voltage power modules by investigating their thermal and electrical properties at different temperatures. The cross-sectional morphology of the prepared materials is characterized by field emission-scanning electron microscopy (FE-SEM). The thermal analysis consists of differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA). The electrical properties, such as dc conductivity, dielectric spectroscopy, thermally stimulated discharge current (TSDC), space charge distribution, and dc breakdown strength, are measured. The molecular orbital energy levels of the EP and SiCw monomers are calculated by quantum chemical calculation (QCC) to analyze the effect of EP interfacial trap sites on the electrical properties. The effect of the EP/SiCw interfacial charges and electrical field distortion on the electrical properties of the EP/SiCw composites with different filler content is investigated by the finite-element method (FEM) simulation. It is found that the 2 wt% EP/SiCw composite has a better thermal performance. The 1 wt% has lower real permittivity, lower trap density, lower space charge accumulation, and higher breakdown strength. The EP/SiCw composites have a higher trap level than the pure EP. The nonlinearity percolation threshold of the EP/SiCw composites is 3 wt%. As the filler content increases, the nonlinear conductivity coefficient of the EP/SiCw composites becomes higher, whereas the breakdown strength decreases due to the electric field distortion near the electrodes caused by the apparent hetero charge accumulation at high filler content. [ABSTRACT FROM AUTHOR]

This work focuses on the performance of epoxy microcomposites, nanocomposites, and micro–nano hybrid composites as the packaging material of the power modules. The experiments consist of thermal conductivity, differential scanning calorimetry (DSC), space charge, dc electrical conductivity, thermally stimulated depolarization current (TSDC), and dielectric spectroscopy measurements. Moreover, the ac stationary electric field simulation and the dc transient electric field simulation are performed to analyze the electrical performance of the epoxy composites as the packaging materials in the power module structure under ac voltage and dc voltage, respectively. The ac field simulation uses dielectric spectroscopy data to obtain the electric field and power loss of the epoxy composites at different temperatures and frequencies under the ac voltage. On the other side, the dc electrical conductivity and TSDC experiment data are used for the dc transient electric field simulation to obtain the electric field considering the space charge under the dc voltage. The results show that the optimal addition of the filler content (20 wt% micro-AlN and 5 wt% nano-AlN) in the epoxy micro–nano hybrid composites can reduce the power loss and electric field at 150 °C and 1 MHz, under the ac condition. Furthermore, considering the space charge under the dc voltage, the electric field can also be reduced in epoxy micro–nano hybrid composites compared to pure epoxy. [ABSTRACT FROM AUTHOR]

This paper focuses on the potential of epoxy resin/aluminum nitride (EP/AIN) nanocomposites through experiments and simulation for the packaging material of a 15 kV SiC-IGBT module. For this purpose, space charge, AC breakdown, thermal conductivity and glass transition temperature $(T_{\mathrm{g}})$ are measured. The electric stress reduction effect in EP/AIN nanocomposites is compared to the traditional packaging material at high-frequency AC and square voltages at 150 °C using the electric field simulation. The experimental results show that the 2 wt% EP/AIN nanocomposite exhibits the least space charge accumulation, highest AC breakdown strength and $T_{\mathrm{g}}$ amongst the tested nanocomposites. The offset of metallization layers plays a crucial role in the impact of the electrical properties of packaging materials on the electric field of the packaging structure. When the offset of metallization layers is zero, the maximum electric stress in the EP/AIN nanocomposites is lower than the traditional packaging material, both under the AC and square voltages. In the EP/AIN nanocomposites, as the frequency or the nano-AIN content increases, the maximum electric stress decreases under the AC voltages. Traditional packaging materials have a steeper transient electric field variation than EP/AIN nanocomposites, although the transient process is smoother at higher frequencies under square voltages for all the considered materials. A simplified physical model is proposed to explain the electric field distribution variation in the packaging structure under the AC voltages. [ABSTRACT FROM AUTHOR]

Well-annotated training samples show necessity in achieving high performance of object detection, but collection of massive samples is extremely laborious and costly. Recently, cut-paste based methods show the potential to augment the training samples by cutting the foreground instances and pasting them on some background regions. However, existing cut-paste based methods hardly guarantee the quality of synthetic images due to lack of mechanism to ensure rationality of the pasted instances (e.g., context, geometry and diversity), limiting the effectiveness of data augmentation. To overcome above issues, this paper proposes a novel Constrained Online Cut-Paste (COCP) method, making an attempt to effectively and efficiently augment training data for improving performance of object detection. Specifically, our COCP generates synthetic images by switching instances of same class from various image pairs in each training mini-batch, ensuring context coherence between the cut instances and the pasted backgrounds. Furthermore, two constraints based on geometric consistency and sample diversity are developed to eliminate counterproductive and meaningless switched instances those suffer from significant geometric discrepancy or lack variations, further improving quality of the synthetic images. The experiments are conducted on both MS COCO and PASCAL VOC datasets using various state-of-the-art detectors (e.g., Faster R-CNN, RetinaNet, FCOS and Mask R-CNN). The results show that our proposed COCP can be well generalized to various datasets and detectors with clear performance gains, while performing favorably against its counterparts. [ABSTRACT FROM AUTHOR]

Compared with global average pooling in existing deep convolutional neural networks (CNNs), global covariance pooling can capture richer statistics of deep features, having potential for improving representation and generalization abilities of deep CNNs. However, integration of global covariance pooling into deep CNNs brings two challenges: (1) robust covariance estimation given deep features of high dimension and small sample size; (2) appropriate usage of geometry of covariances. To address these challenges, we propose a global Matrix Power Normalized COVariance (MPN-COV) Pooling. Our MPN-COV conforms to a robust covariance estimator, very suitable for scenario of high dimension and small sample size. It can also be regarded as Power-Euclidean metric between covariances, effectively exploiting their geometry. Furthermore, a global Gaussian embedding network is proposed to incorporate first-order statistics into MPN-COV. For fast training of MPN-COV networks, we implement an iterative matrix square root normalization, avoiding GPU unfriendly eigen-decomposition inherent in MPN-COV. Additionally, progressive $1\times 1$ 1 × 1 convolutions and group convolution are introduced to compress covariance representations. The proposed methods are highly modular, readily plugged into existing deep CNNs. Extensive experiments are conducted on large-scale object classification, scene categorization, fine-grained visual recognition and texture classification, showing our methods outperform the counterparts and obtain state-of-the-art performance. [ABSTRACT FROM AUTHOR]

Although maximum mean discrepancy (MMD) has achieved great success in unsupervised domain adaptation (UDA), most of existing UDA methods ignore the issue of class weight bias across domains, which is ubiquitous and evidently gives rise to the degradation of UDA performance. In this work, we propose two improved MMD metrics, i.e., weighted MMD (WMMD) and class-specific MMD (CMMD), to alleviate the adverse effect caused by the changes of class prior distributions between source and target domains. In WMMD, class-specific auxiliary weights are deployed to reweigh the source samples. In CMMD, we calculate the MMD for each class of source and target samples. Since the class labels of target samples are unknown for UDA problem, we present a classification expectation-maximization algorithm to estimate the pseudo-labels of target samples on the fly and update the model parameters using estimated labels. The proposed methods can be flexibly incorporated into deep convolutional neural networks to form WMMD and CMMD based domain adaptation networks, which we called WDAN and CDAN, respectively. By combining WMMD with CMMD, we present a CWMMD based domain adaptation network (CWDAN) to further improve classification performance. Experiments show that, both WMMD and CMMD benefit the classification accuracy, and our CWDAN can achieve compelling UDA performance in comparison with MMD and the state-of-the-art UDA methods. [ABSTRACT FROM AUTHOR]

Modeling images/videos with covariance matrices has attracted increasing attentions in various vision tasks, especially in visual classification. For covariances-based visual classification, measuring the distances between covariances is one of the key issues and has been studied for decades. Since the space of covariances is a Riemannian manifold, the geometrical structure of covariances should be favorably considered when designing distance metrics. Although this problem has been widely studied, designing an effective and efficient metric between high-dimensional covariances (HDCOV) for scalable classification is still an open problem. In this paper, we present an information geometry-based distance (IGBD) to tackle this challenge from the perspective of information geometry. Our idea is based on the fact that each covariance can be viewed as a zero-mean Gaussian distribution, and thus the distances between covariances are measured by those between the corresponding Gaussian distributions. The core of our method is to project each distribution, in the form of a set of random samples, to a vector on the tangent space of a common, known distribution on the statistical manifold, based on Fisher information metric and maximum likelihood method. On the tangent space, the Euclidean norm can be used to measure the distances between those sets of projection vectors (or equivalently distributions). The proposed IGBD for HDCOV is computationally efficient and easily combined with a linear support vector machine, suitable for scalable visual classification. The experiments are conducted on various kinds and sizes of benchmarks, and results show the proposed method is efficient and the combination of HDCOV can achieve very competitive performance. [ABSTRACT FROM AUTHOR]