---
license: mit
library_name: coreml
pipeline_tag: image-segmentation
tags:
  - coreml
  - core-ml
  - ios
  - macos
  - apple
  - on-device
  - face-parsing
  - semantic-segmentation
  - bisenet
  - arxiv:1808.00897
---

# Face Parsing — Core ML

*Facial Segmentation, 2019*

Semantic face parsing into 19 regions: skin, nose, eyes, eyebrows, ears, mouth, lip, hair, hat, eyeglass, earring, necklace, neck, cloth, background. 512×512 input.

<p><img src="https://huggingface.co/mlboydaisuke/Face-Parsing-CoreML/resolve/main/media/19082e4ae1.png" alt="Face Parsing demo" width="49%"> <img src="https://huggingface.co/mlboydaisuke/Face-Parsing-CoreML/resolve/main/media/4ca1b439ca.png" alt="Face Parsing demo" width="49%"></p>

Core ML conversion of [zllrunning/face-parsing.PyTorch](https://github.com/zllrunning/face-parsing.PyTorch) for on-device inference on iPhone, iPad and Mac. Converted with `coremltools`; the packages are stateless, so all sequencing and buffering lives in your Swift code.

| | |
|---|---|
| Task | image segmentation |
| Upstream | [zllrunning/face-parsing.PyTorch](https://github.com/zllrunning/face-parsing.PyTorch) |
| Packages | 1 |
| Download size | 47 MB |
| Minimum iOS | 17.0 |
| Peak RAM | ~300 MB |

## Files

| File | Size | Compute units | SHA-256 |
|---|---:|---|---|
| `FaceParsing.mlpackage.zip` | 47 MB | `all` | `a6dd498bb4e19df1…` |
| **Total** | **47 MB** | | |

`compute_units` is not a suggestion -- it is the configuration the conversion was verified against. Moving a package to a different compute unit can silently change the numerics (FP16 attention overflow) or crash on the GPU.

## Download

```bash
hf download mlboydaisuke/coreml-zoo --include "faceparsing/*" --local-dir ./face_parsing
unzip './face_parsing/faceparsing/*.zip' -d ./face_parsing
```

## Use in Swift

```swift
import CoreML

let config = MLModelConfiguration()
config.computeUnits = .all   // as converted — see the table above

// Unzip the .mlpackage, drop it into your Xcode target and Xcode compiles it
// at build time:
let model = try FaceParsing(configuration: config)

// ...or compile a downloaded .mlpackage at runtime:
let compiled = try await MLModel.compileModel(at: mlpackageURL)
let model = try MLModel(contentsOf: compiled, configuration: config)
```

## Demo

- **Sample app** — [CoreML-Face-Parsing](https://github.com/john-rocky/CoreML-Face-Parsing), a standalone iOS project.
- **Models Zoo** — this model is downloadable and runnable inside the [Models Zoo app](https://apps.apple.com/app/id6762083207) on the App Store, no build required.

## Conversion

- Pitfalls hit during conversion (FP16 overflow, ANE buffer limits, stride handling): [`docs/coreml_conversion_notes.md`](https://github.com/john-rocky/CoreML-Models/blob/master/docs/coreml_conversion_notes.md)
- Model index: [CoreML-Models](https://github.com/john-rocky/CoreML-Models)

## License

The conversion inherits the upstream license: **MIT**.

## Credits

- Upstream authors: [zllrunning/face-parsing.PyTorch](https://github.com/zllrunning/face-parsing.PyTorch), 2019
- Core ML conversion: john-rocky (Daisuke Majima)