Skip to content

Official implementation of our ICRA'22 paper: SPIN Road Mapper: Extracting Roads from Aerial Images via Spatial and Interaction Space Graph Reasoning for Autonomous Driving

License

Notifications You must be signed in to change notification settings

wgcban/SPIN_RoadMapper

Repository files navigation

PWC

PWC

SPIN Road Mapper: Extracting Roads from Aerial Images via Spatial and Interaction Space Graph Reasoning for Autonomous Driving (ICRA'22)

Wele Gedara Chaminda Bandara, Jeya Maria Jose Valanarasu, and Vishal M. Patel

Read Paper: Link

Accepted for presentation at the 2022 IEEE International Conference on Robotics and Automation (ICRA), May 23-27, 2022, Philadelphia (PA), USA.

Overview of proposed SPIN module

We build graphs in two spaces: (a) spatial space and (b) a projected latent interaction space from feature maps. Graph reasoning in spatial space extracts connectivity between the road segments, whereas reasoning over interaction space delineates roads from other topographies. Nodes connected with lines in (a) denote how road segments are modeled to understand connectivity in the spatial space. Regions marked with different colors in (b) denote how different semantics are segregated for better road delineation in the interaction space.

Architecture of proposed SPIN module and SPIN pyramid

The architecture of our proposed method. (a) We perform graph reasoning in both spatial and interaction space. (b) The proposed SPIN pyramid module which performs SPIN graph reasoning at multiple scales 1, 1/2, and 1/4 of original feature map to extract multi-scale long-range contextual information.

Proposed network for road segmentation from aerial images

The input images are first feed forwarded to a feature extractor block followed by a bottleneck consisting of stack of two hourglass modules. Then, the output of bottleneck is passed through a segmentation branch which consists of conv layers, our SPIN pyramid and a final classification layer to get the road segmentation map.

A qualitative comparison between our SPIN Road Mapper and the SOTA methods

Reproducing the results

1. Donwloading the datasets

In this paper we used two publically available road segmentation datasets, namely (1) Massachusetts road dataset, and (2) DeepGlobe dataset.

The Massachusetts road dataset can be downloaded from: Click Here

Complete segmentation network with SPIN module

The main module can be found at modelsstack_module.py/StackHourglassNetMTL_DGCNv4

About

Official implementation of our ICRA'22 paper: SPIN Road Mapper: Extracting Roads from Aerial Images via Spatial and Interaction Space Graph Reasoning for Autonomous Driving

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published