We just released data on crown maps for 100 million trees in the National Ecological Observatory Network (NEON) with information on location, species identify, size, and alive/dead status.
We created this dataset by combining deep learning, remote sensing, and extensive field data to build models that can detect and classify individual trees. Models are almost 80% accurate and can be improved with additional field data collection.