2D Synthetic Training Data for Bacterial Cell Segmentation

Synthetic training datasets created by Georgeos Hardo at the University of Cambridge. The data is used to train models for segmenting bacterial cells under various microscopy conditions. It includes images of B. subtilis and E. coli growing in mother machines, on agar pads, and in microfluidic devices.

Use Cases

• Training segmentation models for B. subtilis in mother machines based on phase contrast images.
• Developing models to segment E. coli on agar pads based on phase contrast and fluorescence images.
• Benchmarking segmentation algorithms for cells growing in microfluidic turbidostats.

Strengths

• Data is synthetically generated for specific, controlled microscopy conditions (e.g., 100x oil, phase contrast).
• Covers multiple bacterial species (B. subtilis, E. coli) and growth environments.
• Author is affiliated with the University of Cambridge.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Row count, file formats, and column-level documentation are unknown.
• Last update date is unknown; freshness unverified.

Provenance

Source

Georgeos Hardo, University of Cambridge, via paperswithcode.

Collection Method

Synthetically generated for training machine learning models.

Time Range

unknown

Freshness

unknown

Geography

unknown