Global Context Networks

TL;DR

This paper introduces GCNet, a lightweight global context network that simplifies the non-local approach by using query-independent global context, achieving comparable accuracy with less computation.

Contribution

The paper proposes a query-independent global context block that reduces computation and parameters while maintaining accuracy, improving upon the non-local network for recognition tasks.

Findings

GCNet outperforms NLNet on major benchmarks.

The global context modeled is nearly the same across different query positions.

The simplified GC block reduces parameters significantly.

Abstract

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.