Datasets:
The dataset viewer is not available for this split.
Error code: RowsPostProcessingError
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
MedicinalLeaf13
Dataset: 5,647 full-resolution RGB JPEG derivatives distributed in 13 class-specific image ZIP archives. The labels are collector-assigned, not independently botanically authenticated. The ZIPs contain the actual image files; the support ZIP contains documentation and reproducibility materials.
Dataset creators and contacts
- Md. Muaj Chowdhury — Daffodil International University (DIU), Bangladesh. Email: muaj15-14389@diu.edu.bd.
- Nazim-E-Alam — American International University-Bangladesh (AIUB), Bangladesh. Email: 22-47047-1@student.aiub.edu. Corresponding contact.
- Sadia Akter Sinthia — Daffodil International University (DIU), Bangladesh. Email: sadia15-14369@diu.edu.bd.
Class archives
| Class label | Images | ZIP archive |
|---|---|---|
| Aloe_Vera | 453 | ML13_Aloe_Vera_v1_QC_FILTERED.zip |
| Amloki | 501 | ML13_Amloki_v1_QC_FILTERED.zip |
| Aparajita | 417 | ML13_Aparajita_v1_QC_FILTERED.zip |
| Lemongrass | 401 | ML13_Lemongrass_v1_QC_FILTERED.zip |
| Mehedi_Pata | 402 | ML13_Mehedi_Pata_v1_QC_FILTERED.zip |
| Mint | 405 | ML13_Mint_v1_QC_FILTERED.zip |
| Moringa | 484 | ML13_Moringa_v1_QC_FILTERED.zip |
| Nayan_Tara | 443 | ML13_Nayan_Tara_v1_QC_FILTERED.zip |
| Nim | 399 | ML13_Nim_v1_QC_FILTERED.zip |
| Pathor_Kuchi | 457 | ML13_Pathor_Kuchi_v1_QC_FILTERED.zip |
| Telakucha | 412 | ML13_Telakucha_v1_QC_FILTERED.zip |
| Thankuni | 455 | ML13_Thankuni_v1_QC_FILTERED.zip |
| Tulsi_Pata | 418 | ML13_Tulsi_Pata_v1_QC_FILTERED.zip |
| Total | 5,647 | 13 ZIP archives |
Repository contents
README.md
MedicinalLeaf13_HF_Public_Support_v3_CC_BY_4_0.zip
archives/
ML13_Aloe_Vera_v1_QC_FILTERED.zip
ML13_Amloki_v1_QC_FILTERED.zip
ML13_Aparajita_v1_QC_FILTERED.zip
ML13_Lemongrass_v1_QC_FILTERED.zip
ML13_Mehedi_Pata_v1_QC_FILTERED.zip
ML13_Mint_v1_QC_FILTERED.zip
ML13_Moringa_v1_QC_FILTERED.zip
ML13_Nayan_Tara_v1_QC_FILTERED.zip
ML13_Nim_v1_QC_FILTERED.zip
ML13_Pathor_Kuchi_v1_QC_FILTERED.zip
ML13_Telakucha_v1_QC_FILTERED.zip
ML13_Thankuni_v1_QC_FILTERED.zip
ML13_Tulsi_Pata_v1_QC_FILTERED.zip
Support ZIP contents: image-level public manifest and SHA-256 checksums; class summaries; technical characterization; sanitized acquisition provenance and provisional tracks; taxonomic crosswalk; quality-control records; publication figures; and three reproducibility Python scripts. See the README_DATASET.md inside the support ZIP for a detailed file-by-file guide and commands.
The dataset derives from 6,331 private source files. After 669 byte-identical redundant copies were removed, 5,662 canonical sources remained. A 390-image overview QC sample was reviewed (130 targeted and 260 stratified random); 15 files were excluded with author approval. Of the 5,647 final released derivatives, 375 were included in the retained portion of the screening sample, whereas 5,272 were not part of the 390-image overview review.
Collection years and recorded dates
Two collectors photographed plant leaves in Bangladesh during 2024, 2025, and 2026, primarily in public gardens. For 4,658 released images, source EXIF DateTimeOriginal records dates from 3 October 2024 to 4 September 2026: 370 images are dated 2024, 2,848 are dated 2025, and 1,440 are dated 2026. Another 100 photographs have a tentative filename-derived date of 15 March 2025, and 889 have no recoverable capture timestamp. The year-specific counts refer to photographs with EXIF dates; EXIF dates were not independently checked against collection logs. Camera-model metadata lists iPhone 8, Xiaomi M2101K6P, and realme 7 Pro, with 989 records lacking a camera-model value.
How to download and verify
Download all 13 class ZIPs from Files and versions → archives/ and extract the support ZIP separately. Use Metadata/final_manifest_5647_v1_2.csv to locate and identify each JPEG. Use Integrity/archive_checksums.csv and Code/verify_release.py to verify files. The 13 ZIP archives collectively contain all 5,647 released JPEGs. The dataset does not include validated training, validation, or test splits.
Known limitations and responsible use
- The 13 botanical class labels are collector-assigned, not independently voucher-verified; Mint and Lemongrass have unresolved species-level assignments.
- Images may include multiple viewpoints of the same physical leaf. Plant, leaf and confirmed capture-session identifiers are not available. The 28 metadata-based acquisition tracks are provisional and do not prove biological independence.
- Source image format and EXIF camera model are associated with class labels; models may learn acquisition cues instead of reliable botanical traits.
- EXIF capture dates are missing for 989 photographs, including 100 records with tentative filename-derived dates. Collection years are reported for the dataset as a whole; the missing individual capture dates are not assigned or estimated. EXIF camera-model metadata does not identify the photographer.
- Released JPEG derivatives have embedded acquisition metadata removed. Private originals, original source paths and sensitive GPS details are not distributed.
- Do not use collector labels as definitive proof of medicinal plant identity, nor interpret image-random-split accuracy as specimen-independent performance.
Citation and licence
Dataset DOI: 10.57967/hf/11431
Repository: MedicinalLeaf13 on Hugging Face
Suggested dataset citation:
Md. Muaj Chowdhury, Nazim-E-Alam, and Sadia Akter Sinthia (2026). MedicinalLeaf13: Full-resolution photographic data for 13 collector-labelled plant classes from Bangladesh [Dataset]. Hugging Face. https://doi.org/10.57967/hf/11431
Dataset licence: Creative Commons Attribution 4.0 International (CC BY 4.0). The dataset creators confirm that the relevant rights holders have authorized release of the photographs under this licence. It permits sharing and adaptation, including commercial use, subject to attribution and the other licence conditions. The licence applies to the released photographic dataset and original documentation/metadata authored for it. It does not automatically apply to third-party material or to the Python scripts, for which no separate open-source licence has yet been specified. The licence text is in LICENSE_DATASET.txt inside the support ZIP.
DOI creator metadata: The dataset repository and DOI registration store their own creator records; changing this README does not change the DOI registry.
Companion article: MedicinalLeaf13: Full-resolution photographic data for 13 collector-labelled plant classes from Bangladesh, manuscript for Data in Brief. No article DOI has been assigned.
- Downloads last month
- 30