PLUS: NASA's new approach to detecting methane, detecting oil spills using AI, and more.
98% accuracy is quite good… I wonder if transfer learning would work as well; say train on global data, including some held-out local data, and then fine tune on the local dataset
98% accuracy is quite good… I wonder if transfer learning would work as well; say train on global data, including some held-out local data, and then fine tune on the local dataset