Skip to content

Dataset · nicoboou

104 thousand microscopic cell images

nicoboou/IDRCell100k

Large set of multi-channel cell images from various microscopes for feature extraction.

What's inside

The collection includes 104,093 biological images obtained across 79 experiments using 7 microscopy types. All pictures have a multi-channel structure and are resized to a uniform 224x224 pixels. Text annotations are not specified in the metadata; the dataset focuses on visual features.

What you can do with it

  • Self-supervised pretraining of encoders for biomedical images
  • Extracting visual cell embeddings in drug discovery tasks
  • Training models to work with multi-channel microscopy
Tasks
Embeddings
Size
100K–1M rows
Downloads this month
1.7M
Likes
18

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("nicoboou/IDRCell100k")