期刊
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
卷 45, 期 6, 页码 6881-6895出版社
IEEE COMPUTER SOC
DOI: 10.1109/TPAMI.2020.3047209
关键词
Convolution; Task analysis; Context modeling; Computer architecture; Visualization; Computational modeling; Feature extraction; deep network; self-attention model; global context; object detection
The non-local network (NLNet) proposes a pioneering approach to capture long-range dependencies within an image by aggregating query-specific global context to query positions. However, empirical analysis reveals that the global contexts modeled by NLNet are almost the same for different query positions. In this paper, a simplified network called global context (GC) block is proposed based on a query-independent formulation, which maintains the accuracy of NLNet with significantly reduced computation.
The non-local network (NLNet) presents a pioneering approach for capturing long-range dependencies within an image, via aggregating query-specific global context to each query position. However, through a rigorous empirical analysis, we have found that the global contexts modeled by the non-local network are almost the same for different query positions. In this paper, we take advantage of this finding to create a simplified network based on a query-independent formulation, which maintains the accuracy of NLNet but with significantly less computation. We further replace the one-layer transformation function of the non-local block by a two-layer bottleneck, which further reduces the parameter number considerably. The resulting network element, called the global context (GC) block, effectively models global context in a lightweight manner, allowing it to be applied at multiple layers of a backbone network to form a global context network (GCNet). Experiments show that GCNet generally outperforms NLNet on major benchmarks for various recognition tasks. The code and network configurations are available at https://github.com/xvjiarui/GCNet.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据