LibreEoMT Segmentation
Collection
EoMT (Encoder-only Mask Transformer, DINOv2 backbone) semantic and instance segmentation weights for LibreYOLO. • 6 items • Updated
LibreEoMTl-seg-1280
EoMT-L (Encoder-only Mask Transformer, DINOv2-Large backbone) for COCO 80-class instance segmentation at 1280 px, the high-resolution variant, repackaged for LibreYOLO.
Source
Derived from tue-mps/eomt, checkpoint
tue-mps/coco_instance_eomt_large_1280.
Copyright (c) 2025 Mobile Perception Systems Lab at TU/e. Licensed under the MIT License.
The architecture uses a DINOv2 Vision Transformer backbone (Meta Platforms, Inc., Apache-2.0), fine-tuned end-to-end by the upstream authors and released under MIT.
Modifications
State-dict key remapping only. Learned parameters are unchanged.
See weights/convert_eomt_weights.py in the
LibreYOLO source repository.
Usage
from libreyolo import LibreYOLO
model = LibreYOLO("LibreEoMTl-seg-1280.pt")
results = model.predict("image.jpg")
Install the EoMT runtime dependency with:
pip install "libreyolo[eomt]"
License
MIT License. See the LICENSE and NOTICE files in this repository.
- Downloads last month
- 27