Tree Species Classification

The field studies how to identify tree species from bark, wood, canopy images, hyperspectral and laser-scanning data, and satellite time series, supporting forest inventories, biodiversity monitoring, and urban tree management. Common approaches include convolutional networks and vision transformers, temporal models for seasonal signals, multimodal sensor fusion, and self-supervised or foundation-model features; ecological priors and domain adaptation can help when labels are scarce. Benchmarks increasingly test fine-grained and long-tailed classes across settings such as under-canopy imagery, UAV views, and mixed forest stands. A central challenge is generalizing across species, seasons, sensors, and locations, while evaluations that separate individual trees and validate maps in the field provide more realistic estimates of performance.