Skip to content

Dataset · microsoft

Biomedical images with masks for nine modalities

microsoft/BiomedParseData

Biomedical image dataset with target text descriptions and binary masks for segmentation, detection, and recognition.

What's inside

Official data repository for a paper on a foundation model that jointly segments, detects, and recognizes biomedical objects across nine modalities. The authors processed public segmentation datasets and released a subset in ZIP files. Each example is a 1024x1024 PNG, a list of segmentation target text descriptions, and a binary ground-truth mask.

What you can do with it

  • Train or fine-tune a biomedical image segmentation model
  • Verify object detection and recognition based on text descriptions
  • Create a LoRA or adapter for a multimodal biomedical model
Downloads this month
624
Likes
74
Access
on request on Hugging Face

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("microsoft/BiomedParseData")