Downloads · 30 days
0
Herbivora
mariosandovalmx/Herbivora
Herbivora is a image segmentation model from mariosandovalmx. Use it for the image segmentation task on the model card, and read the license before you ship it in a product. It is set up for pytorch. The card lists the license as other.
Trained checkpoints for Herbivora (or your local clone): contour / ROI completion and herbivory damage segmentation.
Access
Public
Updated Sep 22, 2026
Repo size
685 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.pth587 MB · 100%
At a glance
- Task
- Image Segmentation
- Library
- pytorch
- License
- other
- Access
- Public
- Created
- Jul 28, 2026
- Updated
- Sep 22, 2026
- SHA
- f4807de0
Model card
From the Hugging Face model README
Herbivora — leaf herbivory U-Net weights
Trained checkpoints for Herbivora (or your local clone): contour / ROI completion and herbivory damage segmentation.
Files
| File | Role |
|---|---|
best_unet_shape.pth | Contour U-Net default (mask-to-mask, 512 px; Auto mode fallback) |
best_unet_shape_smooth.pth | Contour specialist — entire / smooth margin (fine-tuned for large C-shaped bites and small edge nicks) |
best_unet_shape_serrated.pth | Contour specialist — serrated margin |
best_unet_shape_lobed.pth | Contour specialist — lobed margin |
best_unet_shape_compound.pth | Contour specialist — compound leaves |
best_model.pth | Damage U-Net (herbivory analysis) |
Not included
- MobileSAM (
mobile_sam.pt) — third-party weights (Apache-2.0). Herbivora downloads them from Ultralytics assets viadownload_models.py. Original work: ChaoningZhang/MobileSAM. - BiRefNet_lite — loaded from
ZhengPeng7/BiRefNet_liteon first use.
Usage
python download_models.py
# or
python download_models.py --repo mariosandovalmx/Herbivora
License
These Herbivora-trained weights (best_unet_shape*.pth, best_model.pth) are released under the
PolyForm Noncommercial License 1.0.0
(same terms as the Herbivora application):
- Allowed: noncommercial research, education, and similar noncommercial use.
- Not allowed without prior written permission: commercial use (selling products/services, commercial workflows, redistribution for commercial purposes).
- Attribution required: if you use these weights in a publication, thesis, or presentation, cite Herbivora (see
CITATION.cffin the GitHub repository).
Full terms: LICENSE in this repository and in the Herbivora GitHub project.
Third-party components (MobileSAM, BiRefNet) remain under their own licenses.
Citation
If you use these weights, please cite the Herbivora project and the original papers for any third-party components you also use (MobileSAM, BiRefNet, Segment Anything).
- Task
- Image Segmentation
- Library
- pytorch
- License
- other
- Created
- Jul 28, 2026
- Updated
- Sep 22, 2026