MODNet portrait matting (fp16)

Browser-ready import artifacts for portrait-matting, produced by SkillSafe's reproducible converter (models/ in skillsafe.ai) from a pinned upstream source. Every byte here is derivable from that source plus the recipe below; nothing was edited by hand.

Provenance

Upstream	https://huggingface.co/Xenova/modnet/tree/fa2fa546052fba4c08921230a26cc69a333fca12
Upstream SHA-256 / commit	fa2fa546052fba4c08921230a26cc69a333fca12
Recipe	recipes/modnet.yaml — sha256 503d08383ef3e8588a9ab8b632eae7800778fbcff86eb27bde3a884a36e3ddeb
Toolchain	Python 3.12.13, torch 2.10.0, onnx 1.23.0, onnxruntime 1.30.0 on Darwin 25.6.0 arm64
Converted	2026-09-22T21:39:27+00:00

Files

file	class	size	SHA-256
config.json	bundle	0.00 MB	e144d8af9b1f09649785c77f592a76bbc69504ae02e43700663b2a9f00d9c8a2
onnx/model_fp16.onnx	registry (fp16)	12.38 MB	25f165da9bfd30830a575f1f0490f1acd995975cb349bc02f3d79332e1fe5cf6
preprocessor_config.json	bundle	0.00 MB	07d83634b1fdd20142ca6e3fe55ab92b558f56d1b0f005ff3a7926f1c9e1165d

registry files are parameter files served from models.skillsafe.ai once vetted; bundle files ship inside an app; registry-shared is a runtime library reused by every model of the same architecture.

Verification

Imported as published upstream (no conversion). Each file is pinned by SHA-256 to its source; every ONNX file passed onnx.checker and a CPU smoke run under onnxruntime with zero-filled inputs at the declared shapes:

file	inputs	outputs	ms
onnx/model_fp16.onnx	input[1, 3, 512, 512]	output[1, 1, 512, 512]	53.6

Use in the browser

import * as ort from "onnxruntime-web";
const session = await ort.InferenceSession.create("https://huggingface.co/skillsafe-ai/modnet/resolve/main/onnx/model_fp16.onnx", { executionProviders: ["webgpu", "wasm"] });

Contract (onnx/model_fp16.onnx): input input float32 ['batch_size', 3, 'height', 'width'] → output output float32 ['batch_size', 1, 'height', 'width']. Opset 11.

Licence and attribution

MODNet: Zhanghan Ke et al. (City University of Hong Kong), Apache License 2.0 (https://github.com/ZHKKKe/MODNet); ONNX export by Xenova (https://huggingface.co/Xenova/modnet).

Licence: Apache-2.0 — notice: https://github.com/ZHKKKe/MODNet/blob/master/LICENSE. The conversion recipe and this model card are part of the SkillSafe repository and carry its licence; the weights remain under the upstream licence above.

The full manifest.json in this repo records the recipe, sources, toolchain (including the uv.lock hash) and per-file verification numbers.