AI Tree Survey
Please follow this link to an interactive map
it may take a few seconds to load
This example demonstrates an analysis pipeline that harnesses drone technology and artificial intelligence to map and measure woodland at the individual tree level. It is suitable for areas where sub-dominant trees are unlikely to be obscured by the canopies of overstorey trees.
Please get in touch to check availability.
Analysis details
A series of RTK corrected images were collected and processed to create a 2cm aerial orthomosaic, 3D point cloud and Digital Surface Model (DSM). A Canopy Height Model (CHM) was derived from the DSM using terrain information from existing LiDAR data. Individual trees were identified in the orthomosaic using a custom Convolutional Neural Network (CNN) deep learning model. Measures to improve the detection of small objects were included during inference. Top height for each tree was extracted from the CHM, with crown segmentation handled by a separate semi-supervised model. Stocking density and summary statistics were calculated within hexagonal plots. An interactive web map presents the final results to users.
Different approaches to estimate biomass and carbon are currently being tested; height-crown allometry is used here. I’m also working on terrestrial point cloud integration to enable diameter at breast height (DBH) measurement and improve understorey coverage.
Woodland Suitability
Please follow this link to an interactive map (desktop only)
it may take a few seconds to load
This example demonstrates a spatial ‘map based’ approach to obtaining and visualising woodland suitability predictions from the Forest Research “Ecological Site Classification (ESC)” Decision Support System.
Analysis details
This approach replaces traditional ‘point’ observations with soil and habitat surveys, leveraging their detailed spatial information to create suitability maps. Tree species and NVC woodland suitability predictions are visualised using an interactive web map, offering a user-friendly platform to asses site sensitivities, constraints and opportunities under different climate scenarios.
