skillsafe-ai/isnet-general-use

IS-Net (DIS) general-use background removal

Browser-ready import artifacts for background-removal, 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://github.com/danielgatis/rembg/releases/download/v0.0.0/isnet-general-use.onnx
Upstream SHA-256 / commit	v0.0.0
Recipe	recipes/isnet-general-use.yaml — sha256 344b7fc0135d28c517f01921c74784fee61d4a31601d80b47d78ad582ada0781
Toolchain	Python 3.12.13, torch 2.10.0, onnx 1.23.0, onnxruntime 1.30.0 on Darwin 25.6.0 arm64
Converted	2026-09-22T18:56:15+00:00

Files

file	class	size	SHA-256
isnet-general-use.onnx	registry	170.37 MB	60920e99c45464f2ba57bee2ad08c919a52bbf852739e96947fbb4358c0d964a

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
isnet-general-use.onnx	input_image[1, 3, 1024, 1024]	output_image[1, 1, 1024, 1024], 1890[1, 1, 1024, 1024], 1891[1, 1, 1024, 1024], 1892[1, 1, 1024, 1024], 1893[1, 1, 1024, 1024], 1894[1, 1, 1024, 1024], input.1252[1, 64, 512, 512], input.1088[1, 64, 256, 256], input.948[1, 128, 128, 128], input.832[1, 256, 64, 64], input.740[1, 512, 32, 32], input.656[1, 512, 16, 16]	1480.6

Use in the browser

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

Contract (isnet-general-use.onnx): input input_image float32 [1, 3, 1024, 1024] → output output_image float32 [1, 1, 1024, 1024], 1890 float32 [1, 1, 1024, 1024], 1891 float32 [1, 1, 1024, 1024], 1892 float32 [1, 1, 1024, 1024], 1893 float32 [1, 1, 1024, 1024], 1894 float32 [1, 1, 1024, 1024], input.1252 float32 [1, 64, 512, 512], input.1088 float32 [1, 64, 256, 256], input.948 float32 [1, 128, 128, 128], input.832 float32 [1, 256, 64, 64], input.740 float32 [1, 512, 32, 32], input.656 float32 [1, 512, 16, 16]. Opset 13.

Licence and attribution

IS-Net / DIS: Xuebin Qin et al., Apache License 2.0 (https://github.com/xuebinqin/DIS). ONNX distributed by rembg (Daniel Gatis, MIT). https://github.com/danielgatis/rembg

Licence: Apache-2.0 — notice: https://github.com/xuebinqin/DIS/blob/main/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.