# fuse-augmentations

> Evidence-backed documentation for a PyTorch tensor-first engine that fuses compatible image augmentation transforms into fewer resampling passes.

Important: this package is not a general drop-in replacement for Kornia, TorchVision, or Albumentations Compose containers. Unknown spatial passthrough operations are unsafe with auxiliary targets. Performance depends on backend, device, shape, batch, and transform mix.

## Start

- [Overview](https://borda.github.io/fuse-augmentations/): Verified capabilities, critical warnings, and navigation by goal.
- [Installation](https://borda.github.io/fuse-augmentations/getting-started/installation/): Base and optional-backend installation.
- [Quickstart](https://borda.github.io/fuse-augmentations/getting-started/quickstart/): Runnable backend-free BCHW tensor example.

## Contracts and guides

- [Capabilities](https://borda.github.io/fuse-augmentations/concepts/capabilities/): Exact backend and transform support boundaries.
- [Backend pipelines](https://borda.github.io/fuse-augmentations/guides/backend-pipelines/): Kornia, TorchVision, Albumentations, native, and mixed-backend usage.
- [Auxiliary targets](https://borda.github.io/fuse-augmentations/guides/auxiliary-targets/): Safe mask, box, and keypoint routing contracts.
- [Reproducibility](https://borda.github.io/fuse-augmentations/guides/reproducibility/): RNG domains and research controls.
- [Known limitations](https://borda.github.io/fuse-augmentations/known-limitations/): Unsafe, approximate, unsupported, and unverified behavior.
- [FAQ](https://borda.github.io/fuse-augmentations/faq/): Direct answers to common developer and researcher questions.

## Research and reference

- [Quality and fidelity](https://borda.github.io/fuse-augmentations/research/quality-and-fidelity/): Why fewer resampling passes can help and why native pixels differ.
- [Benchmarks](https://borda.github.io/fuse-augmentations/research/benchmarks/): Scoped CPU/MPS performance and memory results, including regressions.
- [Methodology](https://borda.github.io/fuse-augmentations/research/methodology/): A reproducible benchmark and parity protocol.
- [Core API](https://borda.github.io/fuse-augmentations/reference/core/): Generated `Compose` and `FusedCompose` API.
- [Configuration API](https://borda.github.io/fuse-augmentations/reference/configuration/): Generated declarative configuration API.
- [Types and targets API](https://borda.github.io/fuse-augmentations/reference/types-and-targets/): Generated enums, descriptors, converters, and coordinate helpers.

## Source

- [GitHub repository](https://github.com/Borda/fuse-augmentations): Source, tests, experiments, and issue tracker.
- [PyPI project](https://pypi.org/project/fuse-augmentations/): Published distributions and installation metadata.