Multimodal Computational Framework for Assessing SDG Contributions

This research elaborates the importance of addressing the United Nations' Sustainable Development Goals (SDGs) related to sustainable consumption and responsible production using multimodal artificial intelligence, computer vision, and machine learning methods. The past hundred years of urban growth in a developing country such as Brazil has been extraordinary, with more than 85% of the population now living in cities. The growing need for land has led countries such as Brazil, Peru and Bolivia to embrace deforestation as a poorlyinformed cure, with disastrous implications on many species residing in the Amazon Rainforest. So we should be good at estimating deforestation levels because more deforestation leads to a lot of really bad environmental issues. We present a multimodal computational pipeline for the multi-label classification of aerial satellite imagery from an open-access dataset covering the region of Brazil's Amazon rainforest. This framework can help in tracking deforestation activities and it may be relevant to other regions globally as well.