The popularization of LiDAR technology, and notably the possibility to multiply acquisition viewpointsthanks to Unmanned aerial vehicles (UAVs) opens-up new opportunities in forest ecology research. Low-cost, unbiased quantitative data on wood volume distribution can now be obtained at different scales (per tree, per tree compartment, per vertical layer, per micro-site etc). Similarly leaf area and leaf angle distribution can be mapped both at individual treeand stand level. High temporal frequency of lidar coverage allowed by UAV systems provides a way of monitoring phenology over time at the individual crown scale. We can now envisage the calibration of architecture/growth models and carbon allocation models for numerous tropical species, while accounting for local biotic interactions andmicroclimatic variations. This will notably provide data for the modelling of carbon and water fluxes and stocks at unprecedented detail and over extents comparable to fluxtower measurements, or to the resolution of current and next-generation space-borne platforms (e.g. Biomass, Gedi). The challenge is now mostly algorithmic, if we are to process large data fluxes of stand level scanning data to decompose tree-level information pertaining to leaves and wood in dense and biogically diverse tropical forests.This poster will summarize the preliminary results of several ongoing studies in French Guyana and Cameroon, endeavouring to obtain (user-controlled) semi-automated quantitative tree structure models, as well as voxel-based estimations of foliage properties over several hectares of forest.