# Local Relation Networks for Image Recognition

## TL;DR

This paper introduces the local relation layer, a novel image feature extractor that adaptively models local pixel relationships, enhancing semantic inference and outperforming traditional convolutional networks on large-scale recognition tasks.

## Contribution

It proposes the local relation layer for adaptive spatial aggregation, improving visual element modeling and semantic inference in image recognition.

## Key findings

- LR-Net outperforms convolutional networks on ImageNet
- Local relation layers improve modeling capacity
- Enhanced semantic inference in image recognition

## Abstract

The convolution layer has been the dominant feature extractor in computer vision for years. However, the spatial aggregation in convolution is basically a pattern matching process that applies fixed filters which are inefficient at modeling visual elements with varying spatial distributions. This paper presents a new image feature extractor, called the local relation layer, that adaptively determines aggregation weights based on the compositional relationship of local pixel pairs. With this relational approach, it can composite visual elements into higher-level entities in a more efficient manner that benefits semantic inference. A network built with local relation layers, called the Local Relation Network (LR-Net), is found to provide greater modeling capacity than its counterpart built with regular convolution on large-scale recognition tasks such as ImageNet classification.