CampanetaWeeds1

The dataset provides pixel-aligned RGB and four-band multispectral imagery acquired in real orchard conditions, enabling cross-spectral weed-detection studies that are scarcely represented in open repositories. Comprehensive, instance-level annotations for six agronomically relevant weed species support fine-grained object detection and class-imbalance investigation in permanent crops. The inclusion of three separate flight acquisitions allows researchers to test temporal generalisation, illumination robustness and domain-adaptation strategies. Accompanying metadata and processing scripts (flight logs, camera parameters, sample notebooks) facilitate reproducibility and rapid experimentation. Down-stream applications span automated weeding, yield protection and decision-support systems, fostering sustainable orchard management and reducing herbicide usage.