Local Relational Networks for Image Recognition

A Pytorch implementation of the local relational layer from Local Relation Networks for Image Recogntion [paper].

Background

This is a unofficial implementation of Local Relation Layer. There has been another implementation of [local-relational-nets][https://github.com/gan3sh500/local-relational-nets] before, but it cant't run when import it. Therefore, we make modification and implement a runable version.

To use the layer:

from local_relation_layer import LocalRelationalLayer

layer = LocalRelationalLayer(channels=64,k=7,stride=1,m=8)
...
output = layer(input)

Note:

Since the implement of 2 x k x k geometric priors is not inferred in paper, we are unaware of how to constuct it and assume that it's a 2 x k x k matrix, one of which denotes offset on x axis and another for offset on y axis. See the code for details.