Land Use Classification using Convolutional Neural Networks Applied to Ground-Level Images
This work provides a novel method for geographic science by enabling land use mapping from ground-level photos, which is not feasible with overhead imagery.
The authors tackled land use classification by using ground-level images from Flickr, addressing challenges like noisy data and achieving over 76% accuracy on an eight-class problem.
Land use mapping is a fundamental yet challenging task in geographic science. In contrast to land cover mapping, it is generally not possible using overhead imagery. The recent, explosive growth of online geo-referenced photo collections suggests an alternate approach to geographic knowledge discovery. In this work, we present a general framework that uses ground-level images from Flickr for land use mapping. Our approach benefits from several novel aspects. First, we address the nosiness of the online photo collections, such as imprecise geolocation and uneven spatial distribution, by performing location and indoor/outdoor filtering, and semi- supervised dataset augmentation. Our indoor/outdoor classifier achieves state-of-the-art performance on several bench- mark datasets and approaches human-level accuracy. Second, we utilize high-level semantic image features extracted using deep learning, specifically convolutional neural net- works, which allow us to achieve upwards of 76% accuracy on a challenging eight class land use mapping problem.