Downloads · 30 days
1.4K
62% of all-time downloads
isnet-general-use-onnx
Ko033/isnet-general-use-onnx
isnet-general-use-onnx is a image segmentation model from Ko033. Use it for the image segmentation task on the model card, and read the license before you ship it in a product. It is set up for transformers.js. The card lists the license as apache-2.0.
ONNX builds of ISNet (general use) from xuebinqin/DIS (Apache-2.0), packaged for Transformers.js.
All-time downloads
2.3K
Public
Repo size
269 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.onnx134 MB · 100%
At a glance
- Task
- Image Segmentation
- Library
- transformers.js
- License
- apache-2.0
- Model type
- custom
- Access
- Public
- Created
- Jul 17, 2026
- Updated
- Jul 19, 2026
- SHA
- 5349b617
Model card
From the Hugging Face model README
isnet-general-use (ONNX, fp16 / q8)
ONNX builds of ISNet (general use) from xuebinqin/DIS (Apache-2.0), packaged for Transformers.js.
- Source ONNX export: the
isnet-general-use.onnxdistributed by the rembg project (MIT); original weights and architecture by Xuebin Qin et al. (DIS, Apache-2.0). - Modifications: the 11 auxiliary graph outputs (side stages d2-d6 and
intermediate feature taps) were removed, keeping only
output_image(stage-1 sigmoid probability map). fp16 and dynamic-int8 (q8) variants were generated with the official Transformers.js conversion scripts.
I/O
- Input
input_image: float32[1, 3, 1024, 1024], RGB, normalized asx / 255 - 0.5(mean 0.5, std 1.0). - Output
output_image: float32[1, 1, 1024, 1024], sigmoid probabilities. Min-max normalize per image for an alpha matte (as done by rembg).
Files
onnx/model.onnx- fp32 (aux outputs stripped)onnx/model_fp16.onnx- fp16onnx/model_quantized.onnx- dynamic int8 (q8)
- Task
- Image Segmentation
- Library
- transformers.js
- Type
- custom
- License
- apache-2.0
- Created
- Jul 17, 2026
- Updated
- Jul 19, 2026