Efficient Deep Semantic Segmentation for Land Cover Classification Using Sentinel Imagery

Abstract

Nowadays, different machine learning approaches, either conventional or more advanced, use input from different remote sensing imagery for land cover classification and associated decision making. However, most approaches rely heavily on time-consuming tasks to gather accurate annotation data. Furthermore, downloading and pre-processing remote sensing imagery used to be a difficult and time-consuming task that discouraged policy makers to create and use new land cover maps. We argue that by combining recent improvements in deep learning with the use of powerful cloud computing platforms for EO data processing, specifically the Google Earth Engine, we can greatly facilitate the task of land cover classification. For this reason, we modify an efficient semantic segmentation approach (U-TAE) for a satellite image time series to use, as input, a single multiband image composite corresponding to a specific time range. Our motivation is threefold: (a) to improve land cover classification performance and at the same time reduce complexity by using, as input, satellite image composites with reduced noise created using temporal median instead of the original noisy (due to clouds, calibration errors, etc.) images, (b) to assess performance when using as input different combinations of satellite data, including Sentinel-2, Sentinel-1, spectral indices, and ALOS elevation data, and (c) to exploit channel attention instead of the temporal attention used in the original approach. We show that our proposed modification on U-TAE (mIoU: 57.25%) outperforms three other popular approaches, namely random forest (mIoU: 39.69%), U-Net (mIoU: 55.73%), and SegFormer (mIoU: 53.5%), while also using fewer training parameters. In addition, the evaluation reveals that proper selection of the input band combination is necessary for improved performance.

A GRASS GIS Scripting Framework for Monitoring Changes in the Ephemeral Salt Lakes of Chotts Melrhir and Merouane, Algeria

“…Despite the need for a standard approach to the classification system, none of the current methods has been accepted as the standard, and many approaches exist. Methods of automated image classification include k-means clustering [109], principal component analysis [110], hierarchical clustering [111], segmentation [112], object-based classification, and deep learning approaches [113]. Among these, clustering uses algorithms that group pixels with common characteristics into clusters that represent different land cover types.…”

Semantic Segmentation with High-Resolution Sentinel-1 SAR Data

“…In ref. [49], a modified U-net with temporal attenuation encoder (U-TAE) was used as a semantic segmentation method. First, the temporal median filter was used to reduce the noise in the images.…”

Annual Past-Present Land Cover Classification From Landsat Using Deep Learning for Urban Agglomerations

“…While these methods achieve high performance on a regional scale, they suffer from limited generalization capability when processing data from multiple urban agglomerations due to their pixel-level operation. This issue can be mitigated using deep-learning, which has been applied in some related works; however, the targets of these works are limited either spatially 8) or temporally 9) . This study aims to (1) develop deep-learning models for 30-m land cover classification in urban agglomerations on a global scale and (2) demonstrate their potential as baselines for long-term analysis of urban land cover changes, particularly temporary or latest…”