SAS Data for Good
SAS Data for Good
SAS Data for Good initiative that leverages data analytics and artificial intelligence to address pressing humanitarian challenges such as poverty, health, human rights, education, and environmental sustainability. Through partnerships with global organizations, SAS applies advanced data science to drive positive social impact and innovation.
SAS validates whale protection program with machine learning:
Data and AI company SAS partnered with Fathom Science, a spin-off from North Carolina State University, to enhance ocean conservation by validating its WhaleCast model, designed to predict the presence of critically endangered North Atlantic right whales. By integrating historical whale sightings with oceanic data, WhaleCast produces heatmaps to hgelp ships avoid whale strike zones. SAS' Data for Good volunteers used SAS Viya, including its Data Maker tool, to generate synthetic data (around 500,000 data points) and trained seven machine learning models. They also added a module to estimate whales' distance from shore. These combined efforts strengthened the model's accuracy, enabling vessels to adapt speed patterns and reduce collision risk.
Galapagos sea turtle recognition:
SAS R&D teamed up with the Galapagos Science Center (a UNC–Chapel Hill and Universidad San Francisco de Quito collaboration) to see if computer vision could identify over 500 individual Galapagos sea turtles from images alone. By merging open-source tools: Python, OpenCV, PyTorch, CVAT, with SAS Viya for preprocessing, modeling, and deployment, the team built an end-to-end pipeline. They collected over 20,000 unlabelled turtle images and used the ConserVision app at SAS conferences to crowd-source labeling. Labeled images were fed into SAS Visual Analytics dashboards to train models capable of predicting a turtle's species, morphotype, individual ID, and a health index. This first-of-its-kind, repeatable process offers conservationists a powerful tool for non-invasive monitoring of endangered species worldwide.