RumexDrone Image Dataset
Fuente:
Zenodo
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Recurso digital |
| Veröffentlicht: |
Zenodo
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866902288607477760 |
|---|---|
| author | Nasser, Hassan-Roland Roland Schrag, Fabian Dionys Sax, Markus Anken, Thomas Stoop, Ralph |
| author_facet | Nasser, Hassan-Roland Roland Schrag, Fabian Dionys Sax, Markus Anken, Thomas Stoop, Ralph |
| contents | <p>RumexDrone is an openly available RGB image dataset for the detection of Rumex in Swiss grasslands. The data were collected in Switzerland between 2020 and 2025 with a DJI Matrice 300 RTK carrying a Zenmuse P1 camera equipped with 35 mm or 50 mm lenses in nadir configuration, typically at about 12 m above ground level using single-shot autofocus. The repository combines four complementary data products designed for different reuse scenarios. AGS_Multiple_Fields contains five fully annotated field subsets totaling 11,155 image tiles of 1024 × 678 pixels. AGS_Multiple_Fields_Embeddings contains 1,513 annotated tiles selected from 50 flights using distances computed on PCA-reduced image embeddings to favor visual diversity. AGS_Multiple_Fields-Flights contains 179 full-resolution flight images of 8192 × 5460 pixels organized by flight date and location. AGS_Multi_Rumex contains randomly selected annotated images from 15 highlighting images at different vegetation stages and light conditions. Across the repository, images and annotations are linked by filename, and the data are intended for object detection, annotation-efficient sampling, benchmark design, domain generalization, and site-specific weed mapping in grassland and pasture systems. The dataset is released through Zenodo under the Creative Commons Attribution 4.0 International license. By packaging raw flight images together with multiple curated annotated subsets, RumexDrone supports direct reuse in common computer vision workflows and methodological comparisons across alternative data selection strategies. </p> <p><span>For more information about our research projects and publications, please visit: </span><a href="https://www.agroscope.admin.ch/agroscope/en/home/topics/economics-technology/smart-farming.html">Agroscope Smart Farming</a></p> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_18662099 |
| institution | Zenodo |
| language | |
| publishDate | 2026 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | RumexDrone Image Dataset Nasser, Hassan-Roland Roland Schrag, Fabian Dionys Sax, Markus Anken, Thomas Stoop, Ralph Computer vision Rumex Rumex/classification Object Detection Deep Learning Drone Unmanned Aerial Vehicle (UAV) Domain Adaption <p>RumexDrone is an openly available RGB image dataset for the detection of Rumex in Swiss grasslands. The data were collected in Switzerland between 2020 and 2025 with a DJI Matrice 300 RTK carrying a Zenmuse P1 camera equipped with 35 mm or 50 mm lenses in nadir configuration, typically at about 12 m above ground level using single-shot autofocus. The repository combines four complementary data products designed for different reuse scenarios. AGS_Multiple_Fields contains five fully annotated field subsets totaling 11,155 image tiles of 1024 × 678 pixels. AGS_Multiple_Fields_Embeddings contains 1,513 annotated tiles selected from 50 flights using distances computed on PCA-reduced image embeddings to favor visual diversity. AGS_Multiple_Fields-Flights contains 179 full-resolution flight images of 8192 × 5460 pixels organized by flight date and location. AGS_Multi_Rumex contains randomly selected annotated images from 15 highlighting images at different vegetation stages and light conditions. Across the repository, images and annotations are linked by filename, and the data are intended for object detection, annotation-efficient sampling, benchmark design, domain generalization, and site-specific weed mapping in grassland and pasture systems. The dataset is released through Zenodo under the Creative Commons Attribution 4.0 International license. By packaging raw flight images together with multiple curated annotated subsets, RumexDrone supports direct reuse in common computer vision workflows and methodological comparisons across alternative data selection strategies. </p> <p><span>For more information about our research projects and publications, please visit: </span><a href="https://www.agroscope.admin.ch/agroscope/en/home/topics/economics-technology/smart-farming.html">Agroscope Smart Farming</a></p> |
| title | RumexDrone Image Dataset |
| topic | Computer vision Rumex Rumex/classification Object Detection Deep Learning Drone Unmanned Aerial Vehicle (UAV) Domain Adaption |
| url | https://doi.org/10.5281/zenodo.18662099 |