Skip to main content
Dryad logo

Data from: Early detection of encroaching woody Juniperus virginiana and its classification in multi-species forest using UAS imagery and semantic segmentation algorithms

Citation

Wang, Lin et al. (2021), Data from: Early detection of encroaching woody Juniperus virginiana and its classification in multi-species forest using UAS imagery and semantic segmentation algorithms, Dryad, Dataset, https://doi.org/10.5061/dryad.9s4mw6mgh

Abstract

Woody plant encroachment into grasslands ecosystems causes significantly ecological destruction and economic losses. Effective and efficient management largely benefits from accurate and timely detection of encroaching species at an early development stage. Recent advances in unmanned aircraft systems (UAS) enabled an easier access to ultra-high spatial resolution images at a centimeter level, together with the latest machine learning based image segmentation algorithms, making it possible to detect small-sized individuals of target species at early development stage and identify them when mixed with other species. However, few studies have investigated the optimal practical spatial resolution of early encroaching species detection. Hence, we investigated the performance of four popular semantic segmentation algorithms (decision tree, DT; random forest, RF; AlexNet; and ResNet) on a multi-species forest classification case with UAS-collected RGB images in original and down-sampled coarser spatial resolutions. The objective of this study was to explore the optimal segmentation algorithm and spatial resolution for eastern redcedar (Juniperus virginiana, ERC) early detection and its classification within a multi-species forest context. To be specific, firstly, we implemented and compared the performance of the four semantic segmentation algorithms with images in the original spatial resolution (0.694 cm). The highest overall accuracy was 0.918 achieved by ResNet with a mean interaction over union at 85.0%. Secondly, we evaluated the performance of ResNet algorithm with images in down-sampled spatial resolutions (1 cm to 5 cm with 0.5 cm interval). When applied on the down-sampled images, ERC segmentation performance decreased with decreasing spatial resolution, especially for those images coarser than 3 cm spatial resolution. The UAS together with the state-of-the-art semantic segmentation algorithms provides a promising tool for early-stage detection and localization of ERC, and the development of effective management strategies for mixed-species forest management.

Methods

Dataset collection:

RGB images were collected at a University of Nebraska-Lincoln rangeland property (41°05'05.2"N, 100°45'53.7"W) located in North Platte, Nebraska. A rotary-wing UAS (Matrice 600 Pro, DJI, Shenzhen, China) with a RGB camera (Zenmuse X5R, DJI, Shenzhen, China) was flown over the study area at an altitude of 30.5 meters above the ground level on May 18th, 2019 with the front and side overlaps of 90% and 85%.

Orthomosaic:

The images were stitched into an orthomosaic in an original spatial resolution of 0.694 cm using Pix4Dmapper software (PIX4D, Lausanne, Switzerland). Geometric correction was performed during the processing by correcting the geolocation of the orthomosaic with those from surveyed by the RTK GNSS system. Four segments were cropped from the orthomosaic with the same size of raw images.

Labelling class name:

The boundary of each tree was manually delineated as polygon shapefile in ArcGIS (ESRI, Redlands, USA). The shapefiles were converted to the raster type with the targeted spatial resolution.

Usage Notes

The following data is included in the dataset:

(1) Raw_Images: 68 raw images (DJI_0xxx.png) and 4 cropped segments from the orthomosaic (DJI_990x.png) in a size of 4608 by 3456 pixels.

(2) Label_Images: labelled images with four classes, redcedar (pixel value as 1), defoliation (pixel value as 2), pine (pixel value as 3), and others (pixel value as 0)

(3) Stitch_Image: stitched orthomosaic.

Funding

McIntire Stennis Fund, Award: 1017851

Hatch Act capacity funding program, Award: 1011130

National Institute of Food and Agriculture, Award: 2018-67007-28529

McIntire Stennis Fund, Award: 1017851

Hatch Act capacity funding program, Award: 1011130