Machine Learning Approaches for Crop Mapping Using Multitemporal Remote Sensing Images Raul Feitosa, PUC-Rio/UERJ
Food security is a major concern worldwide. In this scenario, accurate crop yield estimates are essential for decision makers and planers. For that purpose, satellite remote sensing technology offers repetitive, timely and accurate information of the agricultural activity over large areas at relative low costs. Automatic crop recognition is a challenging task mainly due to the complex spatial and temporal dynamics characteristic of each crop type. This talk presents the most recent Machine Learning based approaches for agricultural mapping using multitemporal satellite images. Methods based on Graphical Models and Deep Learning will be shorty described, their strengths and weaknesses will be discussed and results upon passive (optical) and active (SAR) sensors’ datasets will be compared.