BiRefNet-openvino / README.md
ibrhrsw's picture
model selection
9c0b09e

A newer version of the Gradio SDK is available: 6.14.0

Upgrade
metadata
title: BiRefNet Openvino
emoji: 🏢
colorFrom: pink
colorTo: blue
sdk: gradio
sdk_version: 6.13.0
app_file: app.py
pinned: false
license: mit
short_description: BiRefNet Background Removal on CPU

Runs ibrhr/BiRefNet-openvino-xeon-w2145 OpenVINO models on CPU.

The Space accepts one uploaded image, lets the user choose a model variant, and returns:

  • the predicted foreground mask
  • a transparent PNG with the background removed
  • processing time broken down by preprocessing, inference, and postprocessing
  • runtime specs for the model variant, device, tensor shapes, and image size

Available OpenVINO variants:

  • fp32_1024x1024: FP32, 1024x1024, 4853.2 ms / 0.21 FPS
  • fp32_512x512: FP32, 512x512, 1261.7 ms / 0.79 FPS
  • fp16_1024x1024: FP16, 1024x1024, 4838.1 ms / 0.21 FPS
  • fp16_512x512: FP16, 512x512, 1252.5 ms / 0.80 FPS
  • int8_1024x1024: INT8 NNCF, 1024x1024, 4923.4 ms / 0.20 FPS
  • int8_512x512: INT8 NNCF, 512x512, 1256.7 ms / 0.80 FPS
  • int8wo_1024x1024: INT8 weight-only, 1024x1024, 4972.6 ms / 0.20 FPS
  • int8wo_512x512: INT8 weight-only, 512x512, 1283.3 ms / 0.78 FPS