Models Repository

This repository contains model weights for the SPAD for Vision project.

Model Structure

Models are organized by type:

• dinov3/ - DINOv3 model weights
• flat_surface/ - Flat surface detection models
• material_detection_head/ - Material detection head models
• material_purity/ - Material purity detection models
• spatiotemporal/ - Spatiotemporal detection models
• yolov3/ - YOLOv3 model weights
• yolov8/ - YOLOv8 model weights

Note

Due to GitHub's file size limitations, the actual model files are hosted on Hugging Face Hub:

• Organization: mvplus
• Model repositories: mvplus/dinov3, mvplus/flat_surface, etc.

Usage

Models can be downloaded from Hugging Face Hub using:

from huggingface_hub import snapshot_download
snapshot_download(repo_id="mvplus/dinov3", repo_type="model")

Or via the Hugging Face CLI:

huggingface-cli download mvplus/dinov3