{"success":true,"data":{"author":"Sofia University Faculty of Mathematics and Informatics and SAP Labs Bulgaria","bucketAccessType":"public","bucketName":"su-fmi/sunflower-density-estimation-2025","bucketPrefix":"","category":"Agriculture & Environment","cloudProvider":"huggingface","createdAt":"2026-11-21T22:15:35.923Z","description":"Second season of the Sofia University and SAP Labs Bulgaria sunflower plant density (stand count) dataset, covering a larger field than the 2024 release and adding full-field low-altitude orthophotos as ground truth for scalable density estimation methods, such as spectral characterization from Sentinel-2. Four surveys between 14 May and 17 June 2025 used a DJI Phantom 4 Multispectral drone (blue, green, red, red edge, NIR) flown at 50 m because of a nearby airport. Each survey folder holds raw aerial frames, DJI Terra RGB and per-band orthophotos, NDVI, GNDVI, NDRE, LCI and OSAVI index maps, and low-altitude images at 3, 5 and 10 m for 48 survey points. The low_alt_aerial folder adds orthophotos flown at 14 m and 10 m in June. Files plant_spacing_in_a_row.csv and targeted_plant_density.txt give the sowing design. All imagery is georeferenced GeoTIFF or JPEG (EPSG:4326); about 10,900 files and 174 GB, with a validation script. Useful for plant counting, row detection, multispectral crop monitoring and validating satellite-based density estimates. Metadata states CC BY 4.0 for access and use; credit Sofia University FMI and SAP Labs Bulgaria.","displayTitle":"Sunflower Plant Density Estimation 2025: DJI P4 Multispectral Drone Orthophotos with Full-Coverage Low-Altitude Validation Mosaics, May to June (Sofia University and SAP Labs Bulgaria)","huggingFaceUrl":"https://huggingface.co/datasets/su-fmi/sunflower-density-estimation-2025","id":"363436fc-9871-56ed-bc79-ca21ad5bf046","license":"CC-BY-4.0","licenseEvidenceUrl":"https://huggingface.co/datasets/su-fmi/sunflower-density-estimation-2025","objectCountEstimate":10903,"probeStatus":"ok","referenceUrl":"https://huggingface.co/datasets/su-fmi/sunflower-density-estimation-2025","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"su-fmi/sunflower-density-estimation-2025\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"su-fmi/sunflower-density-estimation-2025\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=su-fmi, license:cc-by-4.0, size_categories:10K<n<100K, downloads=1650"],"format":"mixed","generatedAt":"2026-11-21T22:31:48.020Z","layout":{"example":"plant_spacing_in_a_row.csv","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"plant_spacing\n23\n26\n28\n18\n25\n21\n26\n24\n15\n27\n27\n20\n27\n27\n21\n25\n31\n14\n34\n31\n20\n30\n18\n27\n24\n22\n28\n21\n24\n22\n22","encoding":"utf8","key":"plant_spacing_in_a_row.csv","truncatedAt":512}],"sampleObjects":["02.06.2025/5m/DJI_0040.JPG","02.06.2025/5m/DJI_0041.TIF","02.06.2025/5m/DJI_0042.TIF","02.06.2025/5m/DJI_0043.TIF","02.06.2025/5m/DJI_0044.TIF"],"sidecars":[{"bytes":"---\nlicense: cc-by-4.0\nlanguage:\n- en\npretty_name: Sunflower density estimation dataset from May to June 2025\nsize_categories:\n- 10K<n<100K\n---\n# Dataset Metadata\n\n## Identification Information\n\n### Citation\n\n- **Title**: Sunflower density estimation dataset from May to June 2025  \n- **Originator**: Sofia University - Faculty of Mathematics and Informatics, SAP LABS Bulgaria \n- **Publication Date**: 2025.07.07\n\n### Abstract\nThis dataset is focused on the sunflower plant density estimation problem. It is a continuation of our efforts and \ndataset from 2024. See: [Sunflower density estimation dataset from April to July 2024 ](su-fmi/sunflower-density-estimation-2024). \nIn the current dataset we managed to significantly increase the studied field's area and even more importantly, to include \nvalidation data in the form of low-altitude orthophotos with full field coverage. The latter can help validate the accuracy \nof more scalable estimation algorithms that do not rely on exhaustive plant counting (e.g. spectral characterization from Sentinel-2 data).\n\nDJI P4 Multispectral was again used to collect the dataset and DJI Terra was used for the post processing and reconstruction of orthophotos.\n \n\n### Purpose\n\nThis dataset was developed as part of a research project, investigating the capabilities and application of drones and multispectral cameras for the agricultural domain.\nThe provided data can be used for the following scenarios:\n1) Training models, relying on multispectral data sources.\n2) Improving existing algorithms in the computer vision domain.\n3) Developing and validating methods for sunflower density estimation.\n\n## Time Period of Content\n\n- **Single Date/Time**: Start Date 2025-05-14 to End Date 2025-06-17\n\n## Data Quality Information\n\nComposite images (orthophotos) have been generated with DJI Terra, with 75% frontal and 60% side overlap (50% front and 50% side overlap for the low-altitude orthophotos).\nSome of the surveys have been completed in suboptimal weather conditions (partly cloudy). This resulted in visible variation in color and reflectances in several regions of the orthophotos.\nThe raw data is validated to be complete - representing the entirety of the observed field for every survey. An accompanying validation script is provided with the dataset.\n\n### Horizontal Coordinate System\n\n- **Geographic Coordinate System**: EPSG:4326\n  - **Angular Unit**: Decimal degrees\n  - **Datum**: WGS 84\n  - **Prime Meridian**: Greenwich\n  - **Domain**: Raster\n\n## Entity and Attribute Information\n\n### Detailed Description\n\n#### Entities\nData is organized into directories. Each directory corresponds to one survey and uses **DD.MM.YYYY** format.\n\nEach survey directory contains the following subdirectories:\n- **aerial** - raw aerial footage, used during the reconstruction of the orthophoto with DJI Terra. These photos were taken from an altitude of 50 m, because the field is close to an airport and the altitude of 120 m was not allowed without special permission.\n- **terra** - resulting orthophotos. There are two subdirectories, `default/map` and `lu/map`. The former is a reconstruction with default settings, whereas in the latter, the light uniformity switch was activated.\n  - There is a `result.tif` file, corresponding to the RGB orthophoto and 5 orthophotos for each band, following the `result_<Blue, Green, NIR, Red, RedEdge>.tif` naming pattern.\n  - There are two subdirectories with 5 vegetation index orthophotos, calculated by DJI Terra (GNDVI, LCI, NDRE, NDVI, OSAVI).\n    - **index_map** - these orthophotos contain the vegetation index values in `float32` (range is -1:1)\n    - **index_map_color** - these orthophotos contain a \"false color\" render of the vegetation index values, for the purposes of visualization.\n  - In addition, there are .prj projection file and .tfw georeference file for each orthophoto.\n- **XXm** - where `XX` is either 3, 5 or 10 contains the low-altitude images. For each of the 48 surveying points, there is one RGB image in JPEG and 5 images in TIFF format (corresponding to the 5 bands),\n\nAdditio","encoding":"utf8","key":"README.md","name":"README.md","size":5715,"truncated":true}],"sourceObjectKey":"plant_spacing_in_a_row.csv","starterCell":"import pandas as pd\ndf = pd.read_csv(\"hf://datasets/su-fmi/sunflower-density-estimation-2025/plant_spacing_in_a_row.csv\")\ndf.head()","starterSamplePath":"02.06.2025/5m/DJI_0040.JPG","totalObjects":10903},"summary":"Second season of the Sofia University and SAP Labs Bulgaria sunflower plant density (stand count) dataset, covering a larger field than the 2024 release and adding full-field…","title":"Sunflower Plant Density Estimation 2025: DJI P4 Multispectral Drone Orthophotos with Full-Coverage Low-Altitude Validation Mosaics, May to June (Sofia University and SAP Labs Bulgaria)","updatedAt":"2026-11-22T01:07:16.729Z","visibility":"public","websiteUrl":"https://huggingface.co/datasets/su-fmi/sunflower-density-estimation-2025"}}