|
|
| (2 intermediate revisions by the same user not shown) |
| Line 1: |
Line 1: |
| − | We have several options to test how good BT representations are and to compare them with ground truth.
| |
| | | | |
| − | * Borneo CHM
| |
| − |
| |
| − | * Mantle has a paper on ForestAGB being published in RSE next month; best performance amongst 5 other approaches.
| |
| − |
| |
| − | * Through re.green, we have access to about 670 field plots in Brazil with repeat LiDAR, ALS, and plot data. Felipe Bigliomini works closely with Suzano.
| |
| − |
| |
| − | * Pastis data. This is segmentation task and not suited to pixel-based representations, though.
| |
| − |
| |
| − | * Clement/Mantle has a good crop type dataset for an important agricultural production region in Austria (the so-called ‘Marchfeld’), characterised by a large number of crops in widely different field sizes.
| |
| − |
| |
| − | * David Coomes will make enquiries about access to natural forest, shrubland and grassland ground truth datasets from a global consortium of field plot owners that we can compare with representations (geolocation accuracy may be an issue). Potentially 3 of David’s students will be interested.
| |
| − |
| |
| − | * We might have access to data through NeonScience (a long term NSF-resourced initiative focused on US ecosystem monitoring) https://www.neonscience.org/about
| |
| − |
| |
| − | * Anil, Andres, and Ronita are looking into urban classification datasets. Andrés is assembling a note in the next couple of weeks on what's available and the resolution. Urban mapping is interesting as it could help us explore the wild/urban divide where a lot of expansion is happening (see article in Nature, 2023).
| |
