Model Weights
2,928 Accesses
OmniVision-3D Pre-trained Checkpoints
License
MIT License
Data / File Format
PyTorch / ONNX (4.2 GB)
Access Dataset & Model Weights
Free for non-commercial academic research under MIT License license terms.
Asset Summary & Description
PyTorch model weights pre-trained on 150,000 unannotated volumetric CT & MRI scans for zero-shot medical transfer learning.
Data Specifications & Benchmark Metrics
Model Zoo Details:
- Architectures: ViT-Base (86M params), ViT-Large (304M params).
- Pre-training Objective: Masked Autoencoder (MAE) in 3D spatial voxel grids.
- Frameworks: PyTorch 2.3+, MONAI, ONNX Runtime.
BibTeX Citation:
@dataset{medvisai_omnivision-3d-pre-trained-checkpoints,
title={ OmniVision-3D Pre-trained Checkpoints },
author={MedVis AI Research Laboratory Team},
year={2026},
url={ https://github.com/medvisai/omnivision-3d/releases/tag/v1.0.0 }
}