OpenShot ONNX Exports

This repository builds OpenCV-friendly ONNX model packages for OpenShot and libopenshot. The generated ONNX files and zip packages are static runtime assets: OpenShot should not need Python, PyTorch, or the upstream training frameworks after export.

Model Families

Family	Purpose	Directory
YOLO	Object detection and instance segmentation	yolo/
EfficientSAM	Prompted seed-mask generation for Object Mask	efficient-sam/
Cutie	Video object mask propagation for Object Mask	cutie/

XMem remains in experiments/ as historical scratch work and is not promoted as a supported release family.

Release Artifacts

Generated model binaries are ignored by Git. Build them locally and upload the zip files as GitHub Release assets after reviewing upstream model licenses.

Common outputs:

yolo/models.json          YOLO release catalog
yolo/releases/            YOLO zip packages
efficient-sam/models.json EfficientSAM release catalog
efficient-sam/releases/   EfficientSAM zip packages
cutie/models.json         Cutie release catalog
cutie/releases/           Cutie zip packages

Quick Commands

YOLO:

python yolo/scripts/export_yolo_seg_onnx.py

Cutie:

python cutie/scripts/export_cutie_quality_tiers.py

EfficientSAM:

python efficient-sam/scripts/package_efficient_sam.py

Links

• OpenShot
• OpenShot GitHub
• libopenshot
• OpenCV DNN

Project Notes

This is an export utility, not an official upstream model project. See MODELS.md and NOTICE.md for artifact and licensing notes. The repository code is MIT licensed; upstream model weights and generated ONNX files remain subject to their upstream terms.