Metadata-Version: 2.5
Name: serge
Version: 0.0.1
Summary: SERGE: Seam-Edge Reconciled Grid Ensembling: GPU-native sliced inference for high-resolution object detection (placeholder release)
License-Expression: Apache-2.0
Keywords: computer-vision,object-detection,pytorch,sahi,sliced-inference,small-object-detection
Classifier: Development Status :: 1 - Planning
Classifier: Programming Language :: Python :: 3
Requires-Python: >=3.10
Description-Content-Type: text/markdown

# serge

**SERGE: Seam-Edge Reconciled Grid Ensembling.** GPU-native sliced inference for high-resolution object detection in pure
PyTorch, with seam-aware box fusion.

This is a placeholder release that reserves the name. Install with `pip install serge`, then `import serge`. The first real release is coming soon.

This is an independent project. It is compatible with and inspired by
[SAHI](https://github.com/obss/sahi) (Akyon et al., ICIP 2022), but is not affiliated with it.
