CNNs Fusion for Building Detection in Aerial Images for the Building Detection Challenge

TL;DR

This paper improves building detection in aerial images by fusing multiple CNN segmentation results with a deep combiner, significantly enhancing segmentation accuracy across various cities in the DeepGlobe challenge.

Contribution

It introduces a novel fusion strategy based on a deep combiner that leverages multiple CNN outputs and input data for improved building segmentation.

Findings

Segmentation accuracy improved by up to 7% over baseline.

The fusion method significantly enhances building detection across different cities.

The approach outperforms previous solutions in the challenge.

Abstract

This paper presents our contribution to the DeepGlobe Building Detection Challenge. We enhanced the SpaceNet Challenge winning solution by proposing a new fusion strategy based on a deep combiner using segmentation both results of different CNN and input data to segment. Segmentation results for all cities have been significantly improved (between 1% improvement over the baseline for the smallest one to more than 7% for the largest one). The separation of adjacent buildings should be the next enhancement made to the solution.