# Aquatic Macroinvertebrate Images for miniSASS Stream Biomonitoring (13 Groups)

## Facts

- Provider: GroundTruth and International Water Management Institute (IWMI), CGIAR Initiative on Digital Innovation
- License: CC-BY-4.0 (https://huggingface.co/datasets/CGIAR/aquatic-macroinvertebrate-images)
- Size: 1,302 files
- Format: JPEG (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/CGIAR/aquatic-macroinvertebrate-images
- Access: public, readable without an account
- DOI: https://doi.org/10.7910/DVN/1QQPJ5
- Category: Environmental Science
- Source: https://minisass.org
- Page: https://scigantic.com/archive/3f5969a1-b33b-5704-bc7b-a5c078310e9f
- Keywords: Environmental Science, JPEG, Hugging Face

## About

A photo dataset of freshwater aquatic macroinvertebrates assembled to train machine-learning models that support the mini Stream Assessment Scoring System (miniSASS), a citizen-science biomonitoring method for scoring the ecological condition of streams and rivers in southern Africa. It holds 1,300 images, one hundred for each of the thirteen miniSASS groups, which are identified at roughly Order level (bugs and beetles, caddisflies, crabs and shrimps, damselflies, dragonflies, flat worms, leeches, minnow mayflies, other mayflies, snails, clams and mussels, stoneflies, true flies, and worms), each of which carries a sensitivity score in the index. Annotation was done by R.C.J. and F.J.K. Koen of North-West University; curation by Nicholas Pattinson of GroundTruth with the International Water Management Institute (IWMI) under the CGIAR Initiative on Digital Innovation. Images sit in one folder per group (for example caddisflies/, stoneflies/, snails_clams_mussels/), so they load as an image-folder classification set with Hugging Face datasets or torchvision. The data accompany the IWMI working paper by Koen et al. (2023), 'Digitally improving the identification of aquatic macroinvertebrates for indices used in biomonitoring' (hdl.handle.net/10568/138246), and are also deposited on Harvard Dataverse (doi:10.7910/DVN/1QQPJ5). Licensed CC BY 4.0.

## Load the data

```python
from huggingface_hub import hf_hub_download
path = hf_hub_download(repo_id="CGIAR/aquatic-macroinvertebrate-images", filename="bugs_and_beetles/20220903_133958.jpg", repo_type="dataset")
print(path)
```

## Data structure

File types in sampled files: JPEG

Example file: `bugs_and_beetles/20220903_133958.jpg`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/3f5969a1-b33b-5704-bc7b-a5c078310e9f
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/3f5969a1-b33b-5704-bc7b-a5c078310e9f
- Over MCP: `get_schema_card` with id `3f5969a1-b33b-5704-bc7b-a5c078310e9f` at https://api.scigantic.com/mcp

## Related topics

- [Environmental Science datasets](https://scigantic.com/topics/environmental-science)
- [JPEG datasets](https://scigantic.com/topics/jpeg)
