AqUavplant Dataset: A High-Resolution Aquatic Plant Classification and Segmentation Image Dataset Using UAV

In this AqUavplant dataset, we aim to contribute to the aquatic plant mapping benchmarking datasets for agriculture–computer vision researchers. We captured RGB images using a UAV at low altitudes to acquire higher details of small-sized freshwater aquatic plants in Bangladesh. A realistic demographic and geographic variation is added by collecting images of 31 types of invasive and indigenous aquatic species from nine sites in three locations. There are 197 high-resolution images each containing a lofty number of species. The AqUavplant dataset comprising binary and multiclass semantic segmentation annotations, can enhance data-driven models for automatic aquatic plant mapping. Eventually, the reliable baselines mentioned in the manuscript will aid researchers in leveraging the dataset for improved prediction, monitoring, and obtaining demographic data, thereby advancing common and rare aquatic plant mapping.