Instructions to use jignesh-work/vitaltools-birefnet-lite-512 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers.js
How to use jignesh-work/vitaltools-birefnet-lite-512 with Transformers.js:
// npm i @huggingface/transformers import { pipeline } from '@huggingface/transformers'; // Allocate pipeline const pipe = await pipeline('image-segmentation', 'jignesh-work/vitaltools-birefnet-lite-512');
512x512 fp16 ONNX export of ZhengPeng7/BiRefNet_lite, used by vitaltools.top's Background Remover (Private mode, low-memory devices).
BiRefNet by ZhengPeng7, MIT licence. Weights unchanged; exported from PyTorch at a fixed 512x512 input with a half-precision graph and fp32 input/output.
Export-only changes: deform_conv2d lowered via deform_conv2d_onnx_exporter with static shapes. Max difference vs the fp32 PyTorch model on random input: 2.5e-7.
- Downloads last month
- 122
Model tree for jignesh-work/vitaltools-birefnet-lite-512
Base model
ZhengPeng7/BiRefNet_lite