Searching…
Apr 25, 2019 · The Non-Local Network (NLNet) presents a pioneering approach for capturing long-range dependencies, via aggregating query-specific global context to each query position. However, through a rigorous empi...
Experimental results demonstrate that the proposed GCNet generally outper-forms both non-local networks (with lower FLOPs) and squeeze-excitation networks (with comparable FLOPs).
文章浏览阅读1.5k次，点赞30次，收藏6次。最近的实时语义分割模型，无论是单分支还是多分支，都实现了良好的性能和速度。然而，它们的速度受到多路径块的限制，并且一些模型依赖于高性能的教师模型进行训练。为了克服这些问题，本研究提出了金箍棒网络（GCNet）。具体来说，GCNet在训练过程 ...
图2a是本文的GCNet的上下文建模框架，Context Modeling是采用简化版的Non-local获得空间上的全局信息，Transform是采用SENet捕捉不同通道上的特征，最后的Fusion是通过广播机制将 C*H*W 与 C*1*1 的全局信息按元素相加。
The Non-Local Network (NLNet) presents a pioneering approach for capturing long-range dependencies, via aggregating query-specific global context to each query position. However, through a rigorous empirical analysis,...
a global context modeling frame-work. Then we propose a novel instantiation of this frame-work, the GC block, which is lightweight and can ef ec-tively model long-range dependency. Our GCNet is con-structed via applyi...
Aug 28, 2023 · 文章浏览阅读3.7k次，点赞3次，收藏38次。本文介绍GCNet，一种通过简化非局部网络并融入SE模块的架构，以捕捉全局上下文。实验结果显示，GCNet在保持精度的同时减少计算量，适用于对象检测和图像分类任务。
Nov 17, 2024 · 18、GCNet 论文《GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond》 1、作用 GCNet通过聚合每个查询位置的全局上下文信息来捕获长距离依赖关系，从而改善了图像/视频分类、对象检测和分割等一系列识别任务的性能。
The lightweight property allows us to apply it for multiple layers in a backbone network to construct a global context network (GCNet), which generally outperforms both simplified NLNet and SENet on major benchmarks f...
GCNet: Non-Local Networks Meet Squeeze-Excitation Networks and Beyond Yue Cao, Jiarui Xu, Stephen Lin, Fangyun Wei, Han Hu; Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2019, pp. 0-0...
