Evaluating Sliding Window Multi-Scalar Analysis Importance on Land Use Classification from Satellite Remote Sensing Data

Abstract

Implications of artificial intelligence around the intersection of agricultural technology and satellite remote sensing are only beginning to emerge. One of the application areas in the agriculture sector that can leverage machine learning most immediately is land use classification of remotely sensed data. This research shows that window-based aggregates can help accomplish this task by creating extra layers of information in addition to the original satellite image. Sliding window-based aggregation allows creating these processed data layers efficiently. The results show that adding the processed layers to predictive machine learning models can boost classification accuracy.