metadata
license: apache-2.0
library_name: coreml
pipeline_tag: mask-generation
tags:
- coreml
- core-ml
- ios
- macos
- apple
- on-device
- segment-anything
- sam
- promptable-segmentation
- arxiv:2306.14289
MobileSAM — Core ML
Segment Anything, 2023
Lightweight Segment Anything. Tap any point to generate a segmentation mask. ViT-Tiny encoder + lightweight decoder. ~60× smaller than SAM.
Core ML conversion of ChaoningZhang/MobileSAM 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 | mask generation |
| Upstream | ChaoningZhang/MobileSAM |
| Packages | 1 |
| Download size | 19 MB |
| Minimum iOS | 17.0 |
| Peak RAM | ~300 MB |
Files
| File | Size | Compute units | SHA-256 |
|---|---|---|---|
MobileSAM.zip |
19 MB | all |
0d8d48cb90a48cd8… |
| Total | 19 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 "mobilesam/*" --local-dir ./mobilesam
unzip './mobilesam/mobilesam/*.zip' -d ./mobilesam
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 MobileSAM.zip(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 — SamKit, a standalone iOS project.
- 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.
Credits
- Upstream authors: ChaoningZhang/MobileSAM, 2023
- Core ML conversion: john-rocky (Daisuke Majima)