Comparison of Seedling Detection and Height Measurement Using 3D Point Cloud Models from Three Software Tools: Applications in Forest Restoration

Abstract

A challenge for forest restoration is the monitoring of success, particularly of seedling survivorship. The 3D-point-cloud models, generated from aerial images taken from unmanned aerial vehicles (UAVs), are useful in monitoring vegetation recovery. However, the use of aerial images is challenging due to small seedling size. Many photogrammetry software tools are available for creating 3D-models, but they differ in their performance. This research aims to compare ease of use, seedling detection and the accuracy of seedling height measurement using 3D-point-cloud models from the free versions of three tools: Pix4Dmapper, DroneDeploy, and WebODM. The studied site was at a forest restoration plot of the Siam Cement Group (SCG) Public Company Limited, Lampang, Thailand. The height of 178 planted seedlings was measured manually and used as ground-truth data. On the same day, a UAV was used to automatically capture RGB images of the area, which were processed using the three software tools using their default settings. The percent detection, and the accuracy of height measurements were compared. This study found that DroneDeploy correctly detected 42% of the seedlings, followed by Pix4Dmapper (29%) and WebODM (16%). DroneDeploy and Pix4Dmapper had higher commission error (3%: detection of seedlings that were not there) than WebODM (1%). The differences in seedling detection may arise from differences in algorithms and the default settings used by the software. The accuracy of seedling height measurement was assessed using linear regression against the ground truth data and found that the three software tools performed well (R-squared > 85%). Finally, the pros and cons of each software and their future applications were discussed.