CVPR BiomedSegFM: 3D Biomedical Images for Foundation Model Segmentation

A dataset created for the CVPR 2025 Competition on Foundation Models for 3D Biomedical Image Segmentation. The resource is hosted on Hugging Face by author junma and was last updated on March 18, 2026. It is designed to support tasks in interactive and text-guided 3D segmentation.

Use Cases

• Benchmarking foundation models for 3D biomedical image segmentation based on the competition task settings.
• Developing interactive segmentation tools based on the dataset's 3D biomedical imagery.
• Training text-guided segmentation models based on the described task of text-guided 3D segmentation.

Strengths

• Dataset is associated with a major computer vision conference competition (CVPR 2025).
• Last update timestamp is March 18, 2026, indicating recent maintenance.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

junma on Hugging Face

Collection Method

Created for the CVPR 2025 competition; specific collection method is not detailed.

Freshness

Last updated 2026-03-18 19:24:30