The effectiveness of modeling contextual information has been empirically shown in numerous computer vision tasks. In this paper, we propose a simple yet efficient augmented fully convolutional network(AugFCN) by aggr...
详细信息
The effectiveness of modeling contextual information has been empirically shown in numerous computer vision tasks. In this paper, we propose a simple yet efficient augmented fully convolutional network(AugFCN) by aggregating content-and position-based object contexts for semantic ***, motivated because each deep feature map is a global, class-wise representation of the input,we first propose an augmented nonlocal interaction(AugNI) to aggregate the global content-based contexts through all feature map interactions. Compared to classical position-wise approaches, AugNI is more efficient. Moreover, to eliminate permutation equivariance and maintain translation equivariance, a learnable,relative position embedding branch is then supportably installed in AugNI to capture the global positionbased contexts. AugFCN is built on a fully convolutional network as the backbone by deploying AugNI before the segmentation head network. Experimental results on two challenging benchmarks verify that AugFCN can achieve a competitive 45.38% mIoU(standard mean intersection over union) and 81.9% mIoU on the ADE20K val set and Cityscapes test set, respectively, with little computational overhead. Additionally, the results of the joint implementation of AugNI and existing context modeling schemes show that AugFCN leads to continuous segmentation improvements in state-of-the-art context modeling. We finally achieve a top performance of 45.43% mIoU on the ADE20K val set and 83.0% mIoU on the Cityscapes test set.
The triangle mesh model with huge volume is not conducive to computer storage, analysis and rendering, and the mesh simplification method can effectively reduce the complexity of the triangle mesh model, but there are...
详细信息
Edge computing fulfills the urgent need of users for low-latency and high-quality computing services by transferring tasks from end devices to the edge side of the network for processing through task offloading techni...
详细信息
This paper proposed a new algorithm in the end-to-end automatic speech recognition. For the end-to-end speech recognition model, we select greedy soup instead of the average model parameters in WeNet. We proposed a dy...
详细信息
Achieving high fouling resistance and permeability using membrane separation technology in water treatment processes remains a *** this work,a novel mixed-matrix membrane(MMM)(poly(arylene ether ketone)[PAEK]-containi...
详细信息
Achieving high fouling resistance and permeability using membrane separation technology in water treatment processes remains a *** this work,a novel mixed-matrix membrane(MMM)(poly(arylene ether ketone)[PAEK]-containing carboxyl groups[PAEK-COOH]/UiO-66-NH_(2)@graphene oxide[GO])with superb fouling resistance and high permeability was prepared by the nonsolvent-induced phase separation method,by in-situ growth of UiO-66-NH_(2) on the GO layer,and by preparing hydrophilic *** the basis of the structure and performance analysis of the MMM,the maximum water flux reached 591.25 L·m^(-2)·h^(-1) for PAEK-COOH/UiO-66-NH_(2)@GO,whereas the retention rate for bovine serum albumin increased from 85.40%to 94.87%.As the loading gradually increased,the hydrophilicity of the MMMs increased,significantly enhancing their fouling *** strongest anti-fouling ability observed was 94.74%,which was 2.02 times greater than that of the pure *** the same time,the MMMs contained internal amide and hydrogen bonds during the preparation process,forming a cross-linked structure,which further enhanced the mechanical strength and chemical *** summary,the MMMs with high retention rate,strong permeability,and anti-fouling ability were successfully prepared.
The traditional Monte Carlo rendering method accurately renders 3D scenes but suffers from slow rendering speeds, limiting its suitability for high frame rate applications. Light probes offer a solution for achieving ...
详细信息
Content Aiming at the problem that iris images are easily affected by eyelid and eyelash noise, which leads to low positioning accuracy and poor stability, an iris positioning network based on YOLOv4 model is proposed...
详细信息
In addressing the limited receptive field of single-layer convolution in EEG emotion recognition, the need to stack multiple layers of convolution for expanding the receptive field poses challenges of increased parame...
详细信息
Removing clouds from remote sensing images poses a significant challenge in image analysis. Despite the widespread application of deep learning methods in the field of cloud removal, their ability to extract and integ...
In addressing the issue of unstable boundary adherence in traditional superpixel segmentation algorithms, this paper proposes a novel region-adaptive superpixel segmentation algorithm based on density clustering for c...
详细信息
暂无评论