Remote Sensing of Indicator Plants: Drone Usage in Plant-Based Detection of Chemical Contaminants

Dataset Description This dataset accompanies the paper “Remote Sensing of Indicator Plants” and was used for experiments reported in Section 3.3 (Plant Detection in RGB). The dataset contains UAV-acquired orthomosaic imagery alongside multiple annotation layers and training data for detecting two indicator plant species: oilseed radish (OR) and winter rapeseed (WR). Contents Orthomosaic imageryHigh-resolution RGB orthomosaic used as the primary input for plant detection experiments. Ground truth annotations (Shapefiles)Polygon annotations for individual plants: Oilseed radish (OR) Winter rapeseed (WR) Seeding map (Shapefile)Spatial representation of the experimental seeding layout, indicating where each species was sown. Training dataset (image crops + COCO annotations) Image patches (1024 × 1024 pixels) extracted from the orthomosaic Corresponding annotations in COCO format ⚠️ Important note: Some annotations extend beyond image boundaries. Users must filter or correct out-of-bounds annotations prior to training machine learning models. Usage Notes The dataset is intended for research in UAV-based plant detection, particularly in challenging field conditions with heterogeneous vegetation. Users should ensure proper preprocessing of the COCO annotations before model training. Spatial alignment between orthomosaic, shapefiles, and cropped training data is preserved.