Official Implementation of ”LatentGNN: Learning Efficient Non-local Relations for Visual Recognition“ (ICML 2019)
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.
git clone https://github.com/latentgnn/LatentGNN-V1-PyTorch.git
cd LatentGNN-V1-PyTorch
python setup.py build develop
Please go to maskrcnn-benchmark-latentgnn to use the latentgnn for object detection on COCO dataset.
Please consider citing this project in your publications if it helps your research. The following is a BibTeX reference.
@InProceedings{pmlr-v97-zhang19f,
title = {{L}atent{GNN}: Learning Efficient Non-local Relations for Visual Recognition},
author = {Zhang, Songyang and Yan, Shipeng and He, Xuming },
booktitle = {Proceedings of the 36th International Conference on Machine Learning(ICML)},
pages = {7374--7383},
year = {2019},
volume = {97},
series = {Proceedings of Machine Learning Research},
address = {Long Beach, California, USA},
month = {09--15 Jun},
publisher = {PMLR},
}