# Quadtree Generating Networks: Efficient Hierarchical Scene Parsing with Sparse Convolutions

## TL;DR

Quadtree Generating Networks (QGNs) significantly reduce memory usage in semantic segmentation by hierarchically representing predictions with quadtrees, enabling efficient processing and adaptable inference, demonstrated on multiple datasets.

## Contribution

Introduction of QGNs that use quadtree representations for memory-efficient hierarchical scene parsing in semantic segmentation tasks.

## Key findings

- Achieved over 4× memory reduction compared to large dilated networks.
- Improved mIoU by 3% over similar-memory dilated networks.
- Maintained competitive accuracy with flexible inference schemes.

## Abstract

Semantic segmentation with Convolutional Neural Networks is a memory-intensive task due to the high spatial resolution of feature maps and output predictions. In this paper, we present Quadtree Generating Networks (QGNs), a novel approach able to drastically reduce the memory footprint of modern semantic segmentation networks. The key idea is to use quadtrees to represent the predictions and target segmentation masks instead of dense pixel grids. Our quadtree representation enables hierarchical processing of an input image, with the most computationally demanding layers only being used at regions in the image containing boundaries between classes. In addition, given a trained model, our representation enables flexible inference schemes to trade-off accuracy and computational cost, allowing the network to adapt in constrained situations such as embedded devices. We demonstrate the benefits of our approach on the Cityscapes, SUN-RGBD and ADE20k datasets. On Cityscapes, we obtain an relative 3% mIoU improvement compared to a dilated network with similar memory consumption; and only receive a 3% relative mIoU drop compared to a large dilated network, while reducing memory consumption by over 4$\times$.