Fully convolutional neural networks (FCNs) have shown outstanding performance in many dense labeling problems. One key pillar of these successes is mining relevant information from features layers. However, how to better aggregate multi-level feature maps for salient object detection underexplored. In this work, we present Amulet, a generic aggregating framework detection. Our first integrates into multiple resolutions, which simultaneously incorporate coarse semantics and fine details. Then...

Deep convolutional neural networks (CNNs) have delivered superior performance in many computer vision tasks. In this paper, we propose a novel deep fully network model for accurate salient object detection. The key contribution of work is to learn uncertain features (UCF), which encourage the robustness and accuracy saliency We achieve via introducing reformulated dropout (R-dropout) after specific layers construct an ensemble internal feature units. addition, effective hybrid upsampling...

Deep convolutional neural networks (CNNs) have been successfully applied to a wide variety of problems in computer vision, including salient object detection. To detect and segment objects accurately, it is necessary extract combine high-level semantic features with low-levelfine details simultaneously. This happens be challenge for CNNs as repeated subsampling operations such pooling convolution lead significant decrease the initial image resolution, which results loss spatial finer...

Retinex model-based methods have shown to be effective in layer-wise manipulation with well-designed priors for low-light image enhancement. However, the commonly used handcrafted and optimization-driven solutions lead absence of adaptivity efficiency. To address these issues, this paper, we propose a Retinex-based deep unfolding network (URetinex-Net), which unfolds an optimization problem into learnable decompose reflectance illumination layers. By formulating decomposition as implicit...

Deep networks have been proved to encode high-level features with semantic meaning and delivered superior performance in salient object detection. In this paper, we take one step further by developing a new saliency detection method based on recurrent fully convolutional (RFCNs). Compared existing deep network methods, the proposed is able incorpor- ate prior knowledge for more accurate inference. addition, architecture enables our automatically learn refine map iteratively correcting its...

Deep neural network based methods have made a significant breakthrough in salient object detection. However, they are typically limited to input images with low resolutions (400×400 pixels or less). Little effort has been train networks directly handle segmentation high-resolution images. This paper pushes forward saliency detection, and contributes new dataset, named High-Resolution Salient Object Detection (HRSOD) dataset. To our best knowledge, HRSOD is the first detection dataset date....

Recently, with the advance of deep Convolutional Neural Networks (CNNs), person Re-Identification (Re-ID) has witnessed great success in various applications.However, limited receptive fields CNNs, it is still challenging to extract discriminative representations a global view for persons under non-overlapped cameras.Meanwhile, Transformers demonstrate strong abilities modeling long-range dependencies spatial and sequential data.In this work, we take advantages both CNNs Transformers,...

Visible-Infrared Person Re-Identification (VI-ReID) is a challenging retrieval task under complex modality changes. Existing methods usually focus on extracting discriminative visual features while ignoring the reliability and commonality of between different modalities. In this paper, we propose novel deep learning framework named Progressive Modality-shared Transformer (PMT) for effective VI-ReID. To reduce negative effect gaps, first take gray-scale images as an auxiliary progressive...

In this paper, we propose a novel feature learning framework for video person re-identification (re-ID). The proposed largely aims to exploit the adequate temporal information of sequences and tackle poor spatial alignment moving pedestrians. More specifically, exploiting information, design residual (TRL) module simultaneously extract generic specific features consecutive frames. TRL is equipped with two bi-directional LSTM (BiLSTM), which are respectively responsible describe in different...

Automated segmentation of liver tumors in contrast-enhanced abdominal computed tomography (CT) scans is essential assisting medical professionals to evaluate tumor development and make fast therapeutic schedule. Although deep convolutional neural networks (DCNNs) have contributed many breakthroughs image segmentation, this task remains challenging, since 2D DCNNs are incapable exploring the inter-slice information 3D too complex be trained with available small dataset. In paper, we propose...

Fully convolutional neural networks (FCNs) have shown outstanding performance in many dense labeling problems. One key pillar of these successes is mining relevant information from features layers. However, how to better aggregate multi-level feature maps for salient object detection underexplored. In this work, we present Amulet, a generic aggregating framework detection. Our first integrates into multiple resolutions, which simultaneously incorporate coarse semantics and fine details. Then...

Edge-preserving image smoothing is a fundamental procedure for many computer vision and graphic applications. There tradeoff between the quality processing speed: high usually requires computational cost, which leads to low speed. In this article, we propose new global optimization based method, named iterative least squares (ILS), efficient edge-preserving smoothing. Our approach can produce high-quality results but at much lower cost. Comprehensive experiments demonstrate that proposed...

Video-based person re-identification (Re-ID) aims to automatically retrieve video sequences of the same under non-overlapping cameras. To achieve this goal, it is key fully utilize abundant spatial and temporal cues in videos. Existing methods usually focus on most conspicuous image regions, thus they may easily miss out fine-grained clues due varieties sequences. address above issues, paper, we propose a novel Global-guided Reciprocal Learning (GRL) framework for video-based Re-ID....

Generating a high dynamic range (HDR) image from set of sequential exposures is challenging task for scenes. The most common approaches are aligning the input images to reference before merging them into an HDR image, but artifacts often appear in cases large scene motion. state-of-the-art method using deep learning can solve this problem effectively. In paper, we propose novel convolutional neural network generate HDR, which attempts produce more vivid images. key idea our coarse-to-fine...

Video-based person re-identification aims to associate the video clips of same across multiple non-overlapping cameras. Spatial-temporal representations can provide richer and complementary information between frames, which are crucial distinguish target when occlusion occurs. This paper proposes a novel Pyramid Spatial-Temporal Aggregation (PSTA) framework aggregate frame-level features progressively fuse hierarchical temporal into final video-level representation. Thus, short-term...

Image smoothing is a fundamental procedure in applications of both computer vision and graphics. The required properties can be different or even contradictive among tasks. Nevertheless, the inherent nature one operator usually fixed thus cannot meet various requirements applications. In this paper, we first introduce truncated Huber penalty function which shows strong flexibility under parameter settings. A generalized framework then proposed with introduced function. When combined its...

Abstract As a vital vision task, person re-identification (Re-ID) aims to retrieve the same under non-overlapping cameras. It is very challenging task due presence of complex backgrounds, diverse illuminations and different perspectives. In this work, we integrate advantages convolutional neural networks (CNNs) transformers, propose novel learning framework named multi-level transformer (CMT) for image-based Re-ID. More specifically, first scale-aware feature enhancement (SFE) module extract...

Abstract Global climate change has altered the characteristics of conventional drought events, with an increasing number Slow droughts (SD) rapidly transitioning into Flash (FD). This study introduces a novel multi‐temporal scale identification framework (MTSDIF) that classifies historical agricultural events three types: SD, FD, and Slow‐to‐Flash Drought (SFD). Based on MTSDIF, GLDAS‐Noah root zone soil moisture dataset was used to analyze spatiotemporal characteristics, evolution, driving...