Abstract
Forest inventories can monitor changes in forest carbon sinks and assess forest biodiversity, which is crucial for studying the role of forests in responding to climate change. In recent years, methods for implementing parts of forest surveys on smartphones have often only been supported on specialized devices or a limited number of models. Most methods also assume a large spacing between trees and unoccluded trunks, and hence are unable to cope with realistic forest environments. We present an intuitive mobile application that can run on most common Android devices and estimate tree diameter in near real-time from a single image capture. We collected 154 samples in three countries across varying latitudes, testing our app in challenging conditions including occlusion, leaning trees, and irregular shapes. Our algorithm has a MAE of 2.31 cm and an RMSE of 3.16 cm. In addition to ensuring measurement accuracy, diameter measurement using our app is approximately 5 times faster than traditional manual surveying. Our research offers a low-cost, accurate solution for rapid trunk diameter measurement, expanding its applicability to a wider range of smartphone models and field scenarios.
Supplementary weblinks
Title
Source code for GreenLens app
Description
The source code of our app is publicly available on GitHub at this link.
Actions
View