---
title: finegrain-box-segmenter
canonical_url: "https://www.modelscope.cn/models/cubeai/finegrain-box-segmenter"
md_url: "https://www.modelscope.cn/models/cubeai/finegrain-box-segmenter.md"
repository: cubeai/finegrain-box-segmenter
chinese_name: "图像任意分割(finegrain-box-segmenter)"
last_updated: 2024-11-19
license: "Apache License 2.0"
pipeline_tag: image-segmentation
tasks:
  - image-segmentation
parameters: 94.6M
tensor_type:
  - F16
library_name:
  - safetensors
  - pytorch
frameworks:
  - Pytorch
downloads: 96
stars: 0
---

# finegrain-box-segmenter

> finegrain-box-segmenter - cubeai 在 ModelScope 开源的模型。图像任意分割(finegrain-box-segmenter)

cubeai/finegrain-box-segmenter 是 ModelScope 魔搭社区上的 94.6M 参数image-segmentation模型，采用 Apache License 2.0 许可。

- **Repository**: cubeai/finegrain-box-segmenter
- **License**: Apache License 2.0
- **Tasks**: image-segmentation
- **Parameters**: 94.6M
- **Downloads**: 96
- **Stars**: 0
- **Last updated**: 2024-11-19

Source: https://www.modelscope.cn/models/cubeai/finegrain-box-segmenter

---

# 图像任意分割(finegrain-box-segmenter)

像素级精度图像任意分割。

While building Finegrain, we needed a way to create pixel perfect and high resolution cutouts for objects in images.

The Finegrain Box Segmenter avoids these pitfalls by training MVANet to be a box-promptable High Definition (1024x1024) object-cutout model, making no assumption on what is background and what is foreground: users are fully in control.

## 模型来源

- https://hf-mirror.com/finegrain/finegrain-box-segmenter
- https://github.com/finegrain-ai/refiners

## 使用

https://openi.pcl.ac.cn/cubeai-model-zoo/finegrain-box-segmenter