mlboydaisuke/GFPGAN-CoreML

GFPGAN — Core ML

Face Restoration, 2021

Blind face restoration with generative facial prior. Restores degraded face photos to high quality. 512×512 input/output.

Core ML conversion of TencentARC/GFPGAN 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 to image
Upstream	TencentARC/GFPGAN
Packages	1
Download size	298 MB
Minimum iOS	17.0
Peak RAM	~600 MB

Files

File	Size	Compute units	SHA-256
GFPGAN.mlpackage.zip	298 MB	all	929d0ab30fa739bd…
Total	298 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

hf download mlboydaisuke/coreml-zoo --include "gfpgan/*" --local-dir ./gfpgan
unzip './gfpgan/gfpgan/*.zip' -d ./gfpgan

Use in 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 GFPGAN(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

• Models Zoo — this model is downloadable and runnable inside the Models Zoo app on the App Store, no build required.

Conversion

• Pitfalls hit during conversion (FP16 overflow, ANE buffer limits, stride handling): docs/coreml_conversion_notes.md
• Model index: CoreML-Models

License

The conversion inherits the upstream license: Apache-2.0.

Apache-2.0 except the third-party components (StyleGAN2 / NVIDIA) listed in the upstream LICENSE.

Credits

• Upstream authors: TencentARC/GFPGAN, 2021
• Core ML conversion: john-rocky (Daisuke Majima)