tue-mps/coco_instance_pmt_large_640_dinov3

PMT-DINOv3 (Large, 640px) for COCO Instance Segmentation

Overview

This is the large variant of the PMT-DINOv3 model trained for instance segmentation on COCO at 640x640 resolution.

Model Details

Property	Value
Backbone	DINOv3 ViT-L/16
Input Resolution	640x640
Task	Instance Segmentation
Dataset	COCO

Citation

@inproceedings{cavagnero2026pmt,
    author    = {Cavagnero, Niccolò and Norouzi, Narges and Dubbelman, Gijs and de Geus, Daan},
    title     = {PMT: Plain Mask Transformer for Image and Video Segmentation with Frozen Vision Encoders},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)},
    year      = {2026},
}

Acknowledgements

• Original implementation: tue-mps/pmt
• Paper: arXiv:2503.19108