LatentGNN: Learning Efficient Non-local Relations for Visual Recognition

TL;DR

LatentGNN introduces a scalable graph neural network approach with a latent space to efficiently model non-local relations, significantly improving visual recognition performance while reducing computational complexity.

Contribution

It proposes a novel low-rank GNN with a latent space to efficiently capture non-local dependencies in visual recognition tasks.

Findings

Outperforms prior methods with large margins

Maintains low computational cost

Effective on multiple visual recognition tasks

Abstract

Capturing long-range dependencies in feature representations is crucial for many visual recognition tasks. Despite recent successes of deep convolutional networks, it remains challenging to model non-local context relations between visual features. A promising strategy is to model the feature context by a fully-connected graph neural network (GNN), which augments traditional convolutional features with an estimated non-local context representation. However, most GNN-based approaches require computing a dense graph affinity matrix and hence have difficulty in scaling up to tackle complex real-world visual problems. In this work, we propose an efficient and yet flexible non-local relation representation based on a novel class of graph neural networks. Our key idea is to introduce a latent space to reduce the complexity of graph, which allows us to use a low-rank representation for the graph affinity matrix and to achieve a linear complexity in computation. Extensive experimental evaluations on three major visual recognition tasks show that our method outperforms the prior works with a large margin while maintaining a low computation cost.