Sorghum Panicle Detection Ground-UAV Dataset

This dataset consists of RGB images captured during the early-mid flowering stage of sorghum plots at the University of Queensland's Gatton campus in December 2021. The dataset is divided into two subsets: images acquired using a UAV (Unmanned Aerial Vehicle) and images captured with a ground camera. Each subset includes bounding box annotations for training and evaluating object detection models.

The ground images in this dataset were captured using the OpenCV OAK-1 camera, positioned at a height of approximately 1.5 meters above the canopy. These images have a raw resolution of 2160x2160 pixels and a Ground Sampling Distance (GSD) of approximately 0.030 cm per pixel. The UAV images were acquired using the DJI Matrice 300 drone, flying at an altitude of 20 meters. These images have a raw resolution of 8192x3040 pixels and a GSD of 0.25 cm per pixel. The inclusion of both ground and UAV images with different resolutions and GSDs provides a diverse range of data for training and evaluating object detection models for sorghum panicle detection.

The dataset is associated with the publication titled "From Prototype to Inference: A Pipeline to Apply Deep Learning in Sorghum Panicle Detection", James et al. 2023. The publication can be accessed via the DOI link: https://doi.org/10.34133/plantphenomics.0017