Isnet General Use Onnx

--- license: apache-2.0 library_name: transformers.js pipeline_tag: image-segmentation tags: - onnx - background-removal - dichotomous-image-segmentation --- # isnet-general-use (ONNX, fp16 / q8) ONNX builds of **ISNet (general use)** from [xuebinqin/DIS](https://github.com/xuebinqin/DIS) (Apache-2.0), packaged for [Transformers.js](https://huggingface.co/docs/transformers.js). - Source ONNX export: the `isnet-general-use.onnx` distributed by the [rembg](https://github.com/danielgatis/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 as `x / 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` - fp16 - `onnx/model_quantized.onnx` - dynamic int8 (q8)

Source context: 763 downloads · 0 likes · Pipeline image-segmentation · Library transformers.js · Repo Ko033/isnet-general-use-onnx

Source context: 76 downloads · 0 likes · Pipeline image-segmentation · Library transformers.js · Repo Ko033/isnet-general-use-onnx