GCNet: Non-Local Networks Meet Squeeze-Excitation Networks and Beyond

TLDR

NLNet captures long‑range dependencies by aggregating query‑specific global context, and its simplified design resembles SENet. The authors aim to build a query‑independent, lightweight network that preserves NLNet accuracy while reducing computation. They propose a three‑step global‑context framework instantiated with a lightweight GC block that efficiently models global context across layers. The GCNet, built from this framework, outperforms both simplified NLNet and SENet on major recognition benchmarks, demonstrating that global contexts are largely identical across query positions.

Abstract

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, we have found that the global contexts modeled by non-local network are almost the same for different query positions within an image. 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 observe that this simplified design shares similar structure with Squeeze-Excitation Network (SENet). Hence we unify them into a three-step general framework for global context modeling. Within the general framework, we design a better instantiation, called the global context (GC) block, which is lightweight and can effectively model the global context. 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 for various recognition tasks.