Meta-Album PlantNet: 25 Plant Species Images from Citizen Science

120,688 labeled images of plants across 25 species, sampled from the Pl@ntNet-300k dataset. Images were contributed and verified by citizen botanists worldwide, with each image reviewed by an average of 2.03 users. The dataset was created by Meta-Album authors in March 2022, with original data sourced from the Pl@ntNet Project.

Use Cases

• Training few-shot image classifiers based on the 25 distinct plant species classes.
• Benchmarking meta-learning algorithms on a standardized, pre-processed image dataset.
• Developing models for automated plant species identification from citizen-science imagery.
• Studying label ambiguity and long-tailed distributions in real-world image datasets.

Strengths

• Contains 120,688 total images, providing substantial volume for training.
• Images are standardized to a uniform 128x128 pixel resolution.
• Labels have been verified through a citizen-science process, with an average of 2.03 reviews per image.
• Dataset is focused on 25 populous classes, enabling concentrated study.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Data may reflect geographic or contributor bias inherent to citizen-science platforms.

Provenance

Source

Pl@ntNet Project, via the Pl@ntNet-300k dataset.

Collection Method

Sampled from the Pl@ntNet-300k dataset, with images contributed and verified by citizen botanists.

Freshness

Created Date is 01 March 2022; freshness unverified.

Geography

Global (images sourced by citizen botanists from around the world).