Segmented Vegetation-Density Drone Dataset for Peatland Vegetation Classification

Published: 24 September 2024| Version 1 | DOI: 10.17632/fss67tz8w8.1
Contributors:
Yuslena Sari,
,
,
,

Description

Vegetation Density of Peatland Drone dataset consists of 3 density classes namely bare, softly grazed and heavily grazed. The images were taken in The Liang Anggang Protection Forest Block I in Banjarbaru, South Kalimantan, Indonesia. an area of 479 hectares of block 1 area which is filled with land such as agriculture, plantations, roads and settlements and an area of 494 hectares full of weeds. In addition, the location is also categorized as peat land. The location, especially in block 1, fulfils the characteristics and suitability of the need for data collection for land cover classification in terms of the type of vegetation density that can be seen with the human eye during observations and surveys. Bare/un-grazed typically consists of dry shrubs, soil, water, and settlements, while heavily grazed typically consists of vegetation that has a density between 15 and 30 meters. Meanwhile, softly grazed typically consists of shrubs that are 5 to 6 meters apart and fresh shrubs up to a maximum of 3 meters. Image capture was carried out from November - December 2021 using DJI Mavic Pro drone.The resolution of the acquired image is 4000 x 3000 pixel with total of 300 images, each corresponding class have 100 images and its segmentation label.

Files

Institutions

Universitas Lambung Mangkurat

Categories

Computer Science Applications, Image Segmentation, Peat, Vegetation

Funding

Direktorat Riset Dan Pengabdian Kepada Masyarakat

056/E5/PG.02.00.PL/2024

Universitas Lambung Mangkurat

1014/UN8.2/PG/2024

Licence