Microcolony Domain Adaptation: Phase Contrast Microscopy of Six Foodborne Bacteria under Shifted Imaging Conditions

Provider
Bhattacharya, Wasit, Earles, Nitin and Yi (Food AI Nexus)
License
CC-BY-4.0 (source)
Size
1,205 files
Format
JPEG, JSON Lines (in sampled files)
Stored at
Hugging Face (https://huggingface.co/datasets/food-ai-nexus/microcolony-domain-adaptation)
Access
Free and public. No account with the data provider is needed.
Category
Health & Biomedical
Source
https://huggingface.co/datasets/food-ai-nexus/microcolony-domain-adaptation

About this dataset

Microscopy images of bacterial microcolonies from six species, Bacillus coagulans, Bacillus subtilis, Escherichia coli 1612, Listeria innocua, Salmonella enterica Enteritidis and Salmonella enterica Typhimurium, built to test classifiers when optical and biological conditions change. Released with Bhattacharya, Wasit, Earles, Nitin and Yi (2025), Frontiers in Artificial Intelligence, on adversarial domain adaptation for AI microscopy. The source domain is standard phase contrast at 60x after 3 h incubation (train, 539 images; test_standard, 90 held-out images); target domains vary objective, incubation time and imaging modality (train_fewshot, 150 labeled images for few-shot adaptation; test_ood, 420 out-of-distribution images). Images are JPEG in class subfolders with JSONL metadata per split, about 130 MB in total, and load with datasets.load_dataset. Supports domain adaptation, few-shot learning and robustness benchmarking for rapid detection of foodborne pathogens. Licensed CC BY 4.0 by the authors; cite the Frontiers paper.

How to load this dataset in Python

from datasets import load_dataset
ds = load_dataset("food-ai-nexus/microcolony-domain-adaptation")
print(ds)

Data structure

File types in sampled files: JPEG, JSON Lines

Files

Example file: test_ood/metadata.jsonl

Analyze this dataset without downloading it

On Scigantic, Microcolony Domain Adaptation: Phase Contrast Microscopy of Six Foodborne Bacteria under Shifted Imaging Conditions opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.

This page is also available as Markdown and as JSON.

Related Health & Biomedical datasets