React Native ML Kit — Subject Segmentation

@nitro-mlkit/subject-segmentation · on-device Google ML Kit via Nitro Modules — JSI, no bridge.

⚠️ Beta (0.1.0-beta.x). Android-only. Native path verified on-device; the full run needs the Google Play optional model, which downloads on first use (slow on emulators). See Platform status.

High-performance, on-device subject segmentation for React Native, built with Nitro Modules (JSI, no bridge).

Powered by Google ML Kit. Cuts out the main subject(s) of a photo from the background and returns a transparent PNG cutout. Downloads a small model at runtime on first use, then runs entirely on-device — no network.

Installation

npm install @nitro-mlkit/subject-segmentation react-native-nitro-modules

No config plugin (autolinked Expo module). Install and npx expo prebuild. Not available in Expo Go.

Usage

import { NitroSubjectSegmenter } from "@nitro-mlkit/subject-segmentation";

const { foregroundUri, width, height, subjectCount } =
  await NitroSubjectSegmenter.segment(imageUri);
// foregroundUri -> a PNG cutout of the subject(s) with a transparent
// background; render it, composite it, or save it.

NitroSubjectSegmenter.isAvailable(); // true on Android, false elsewhere

SubjectSegmentationResult

| field | type | notes | | ----- | ---- | ----- | | foregroundUri | string | file:// uri of a transparent PNG cutout of the subject(s) | | width | number | cutout width in pixels | | height | number | cutout height in pixels | | subjectCount | number | number of distinct subjects found |

Notes

• First call downloads a model. ML Kit fetches the subject-segmentation model on first use; that call is slower and needs connectivity. Subsequent calls are fully on-device.
• Beta ML Kit dependency. Uses com.google.mlkit:segmentation-subject (currently a beta artifact).

Platform status

| Platform | Min | Status | | -------- | --- | ------ | | Android | API 26+ | ⚙️ Native path verified (Pixel 9, API 36); full result pending Google Play optional-model download (slow on emulator — validate on a real device) | | iOS | — | ❌ Not supported — Google ML Kit has no Subject Segmentation API on iOS |

isAvailable() returns false on iOS and segment() rejects there, so you can call it safely from cross-platform code and gate on the result.

Part of nitro-mlkit

The full ML Kit suite on Nitro — see the other @nitro-mlkit/* packages.

License

MIT © Gonzalo Polo