Project Abstract

Accurate land cover classification is a vital task for both food security and environmental monitoring, and can be used to measure the distribution of subsidies. In recent years, crop classification has benefited significantly from advancements in satellite technology and more powerful machine learning techniques. Allowing for the large high-resolution multi-spectral images of the earth to be processed, enabling researchers and farmers to monitor land-use and crop health with unprecedented accuracy. Primarily, optical and infer-red data from Sentinel-2 and land-sat are used for crop-classification however, Synthetic Aperture Radar (SAR) data from the Sentinel-1 constellation offers a more frequent and consistent source of land data due to its all-weather and day-night capabilities. The purpose of this study was to evaluate the feasibility of the unsupervised machine learning technique k-means for crop-classification on large time series of Setninel-1 SAR data in the Brandenburg region of Germany. The Sentinel-1 constellation recorded one-hundred and twenty-two snapshots of the region over a one-year period in 2017, each one offering insights into the earth?��s surface in which patterns can be identified using machine learning. Through preprocessing, feature extraction, and clustering, we aim to delineate different land cover types and assess the effectiveness of the proposed methodology. Crop classification happens on a global scale therefore, a lightweight unsupervised machine learning technique was identified, to evaluate its potential use on a wider land area. The findings presented promising results from k-means with effective data-preprocessing, The positives and negatives of the technique are explored and highlighted in the project With the hopes of identifying unique approaches to data pre-processing and k-means implementation.

Keywords: Data Science, Machine Learning, Agriculture

 Conference Details

Session: Poster Session B at Poster Stand 78

Location: Sir Stanley Clarke Auditorium at Wednesday 8th 09:00 – 12:30

Markers: Jay Morgan, Trang Doan

Course: BSc Computer Science, 3rd Year

Future Plans: I’m undecided