Satellite Imagery Feature Detection using Deep Convolutional Neural Network: A Kaggle Competition

TL;DR

This paper presents a deep learning approach for satellite imagery feature detection that achieves high accuracy without complex ensembling, enabling scalable deployment in automatic satellite image analysis systems.

Contribution

The paper introduces a modified fully convolutional neural network tailored for multispectral satellite data and a training pipeline that avoids ensembling, facilitating scalable deployment.

Findings

Achieved third place in Kaggle challenge with competitive accuracy.

Method does not rely on complex ensembling techniques.

Approach is scalable for production deployment.

Abstract

This paper describes our approach to the DSTL Satellite Imagery Feature Detection challenge run by Kaggle. The primary goal of this challenge is accurate semantic segmentation of different classes in satellite imagery. Our approach is based on an adaptation of fully convolutional neural network for multispectral data processing. In addition, we defined several modifications to the training objective and overall training pipeline, e.g. boundary effect estimation, also we discuss usage of data augmentation strategies and reflectance indices. Our solution scored third place out of 419 entries. Its accuracy is comparable to the first two places, but unlike those solutions, it doesn't rely on complex ensembling techniques and thus can be easily scaled for deployment in production as a part of automatic feature labeling systems for satellite imagery analysis.