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

By Vladimir I. Iglovikov, Sergey Mushinskiy, Vladimir Osin. 2017

178 citations recorded by the source corpus.

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.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Media Technology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

Albumentations: fast and flexible image augmentations (2018, methodology): “For example, on of the top-3 performing solutions in the DSTL Satellite Imagery Feature Detection challenge on Kaggle used combinations of random cropping and a random transformation from dihedral group Dih 4 [11].”
Fully Convolutional Network for Automatic Road Extraction from Satellite Imagery (2018, background): “This type of network architecture proven themselves in binary image segmentation competitions for satellite image analyses [6, 7, 8, 15, 16] and other applications [13, 12].”
Angiodysplasia Detection and Localization Using Deep Convolutional Neural Networks (2018, methodology): “2.2 Model architecture and training In this work we evaluate 4 different deep architectures for segmentation: U-Net (Ronneberger et al., 2015; Iglovikov et al., 2017a), 2 modifications of TernausNet (Iglovikov and Shvets, 2018), and AlbuNet34, a modifications of LinkedNet (Chaurasia and…”
Feature Pyramid Network for Multi-class Land Segmentation (2018, background): “The advantages of this data compared to aerial imagery are the almost worldwide availability, and some-times the imagery data contains additional spectral channels [18, 9].”
Automatic Instrument Segmentation in Robot-Assisted Surgery Using Deep Learning (2018, methodology): “By minimizing this loss function, we simultaneously maximize probabilities for right pixels to be predicted and maximize the intersection J between masks and corresponding predictions [9].”
TernausNet: U-Net with VGG11 Encoder Pre-Trained on ImageNet for Image Segmentation (2018, methodology): “izing this loss function, we simultaneously maximize probabilities for right pixels to be predicted and maximize the intersection, J between masks and corresponding predictions. For more details, see [5]. At the output of a given neural network, we obtain an image where each pixel corresponds to a p”
TernausNetV2: Fully Convolutional Network for Instance Segmentation (2018, background): “This loss was inspired by [13] where authors proposed a way to generalize discrete Jaccard index (also known as intersection over union) into a differentiable form.”