Get Started
Home
Topics
Search
Library
Research questionWhen labeled plots are scarce, should tree-species mapping use foundation-model embeddings or engineered multi-temporal features?Tree-species classification from Earth-observation data must distinguish species in both pure and mixed stands. The relative usefulness of learned embeddings and engineered temporal signals may change substantially as labeled plot availability changes.
AI
Computer Vision
Image & Video Processing
Machine Learning
Research Paper
Technology
Latest papersRecent research connected to this question, newest first.Tree species mapping in Denmark: A comparison of spectral-temporal features with geospatial foundation model embeddingsThe evidence concerns Denmark, using National Forest Inventory plots, multi-temporal Sentinel-1 and Sentinel-2 observations, canopy height, and TESSERA or AlphaEarth embeddings. Random forest, XGBoost, and MLP classifiers are assessed for pure and mixed stands, with the best model applied to a 10 m national map; the results do not establish performance beyond this setting.research paper · Sep 4, 2026
Related questions
How can individual-tree species be classified from multispectral airborne laser scans when field references are scarce?How can plant point-cloud segmentation adapt across species and sensing conditions with few labeled examples?How should landmark-derived features preserve task-relevant geometry under nuisance variation in structured image classification?How can compact multimodal Earth-observation models handle missing sensors and changing spatial resolutions?