| --- |
| license: cc-by-nc-sa-4.0 |
| license_link: https://github.com/wyf0912/SinSR/blob/main/LICENSE |
| library_name: coreml |
| pipeline_tag: image-to-image |
| tags: |
| - coreml |
| - core-ml |
| - ios |
| - macos |
| - apple |
| - on-device |
| - super-resolution |
| - diffusion |
| - one-step |
| - arxiv:2311.14760 |
| --- |
| |
| # SinSR — Core ML |
|
|
| *Single-Step Super-Resolution, 2024* |
|
|
| 4× super-resolution via single-step diffusion. 256→1024. Swin Transformer denoiser (FP32). |
|
|
| <p><img src="https://huggingface.co/mlboydaisuke/SinSR-CoreML/resolve/main/media/75b8b5fc68.png" alt="SinSR demo"></p> |
|
|
| Core ML conversion of [wyf0912/SinSR](https://github.com/wyf0912/SinSR) 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 | [wyf0912/SinSR](https://github.com/wyf0912/SinSR) | |
| | Packages | 3 | |
| | Download size | 517 MB | |
| | Minimum iOS | 17.0 | |
| | Peak RAM | ~600 MB | |
|
|
| ## Files |
|
|
| | File | Size | Compute units | SHA-256 | |
| |---|---:|---|---| |
| | `SinSR_Encoder.mlpackage.zip` | 39 MB | `cpuAndGPU` | `fdec09d17561ec1b…` | |
| | `SinSR_Denoiser.mlpackage.zip` | 420 MB | `cpuOnly` | `b31374c2d539b2cd…` | |
| | `SinSR_Decoder.mlpackage.zip` | 58 MB | `cpuAndGPU` | `b8b9a7b52d6b240c…` | |
| | **Total** | **517 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 "sinsr/*" --local-dir ./sinsr |
| unzip './sinsr/sinsr/*.zip' -d ./sinsr |
| ``` |
|
|
| ## Use in Swift |
|
|
| ```swift |
| import CoreML |
| |
| let config = MLModelConfiguration() |
| config.computeUnits = .cpuAndGPU // 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 SinSR_Encoder(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) |
| ``` |
|
|
| > This model is split into 3 Core ML packages that are driven in sequence from Swift. Load them one at a time, copy the outputs out of the `MLMultiArray` buffers and release each model before loading the next — two large Core ML models resident at once will OOM on an iPhone. |
|
|
| ## Demo |
|
|
| - **Sample app** — [`sample_apps/SinSRDemo`](https://github.com/john-rocky/CoreML-Models/tree/master/sample_apps/SinSRDemo), a standalone SwiftUI 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 |
|
|
| - Script: [`convert_sinsr.py`](https://github.com/john-rocky/CoreML-Models/blob/master/conversion_scripts/convert_sinsr.py) |
| - 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: **CC BY-NC-SA 4.0**. |
| See [https://github.com/wyf0912/SinSR/blob/main/LICENSE](https://github.com/wyf0912/SinSR/blob/main/LICENSE). |
|
|
| > Non-commercial, attribution, share-alike. |
|
|
| ## Credits |
|
|
| - Upstream authors: [wyf0912/SinSR](https://github.com/wyf0912/SinSR), 2024 |
| - Core ML conversion: john-rocky (Daisuke Majima) |
|
|
