---
title: MFH
canonical_url: "https://www.modelscope.cn/datasets/OmniData/MFH"
md_url: "https://www.modelscope.cn/datasets/OmniData/MFH.md"
repository: OmniData/MFH
last_updated: 2024-07-04
license: "[MIT]"
storage_size: "535 MB"
domain:
  - publishDate
  - publisher
  - paperUrl
  - publishUrl
  - mediaTypes
  - displayName
tasks:
  - 2021-01-01
  - "Blood Transfusion Hematology Hospital"
  - "https://arxiv.org/pdf/2109.02917.pdf"
  - "https://github.com/willogy-team/hand-gesture-recognition-smc2021"
  - Image
  - MFH
downloads: 65
stars: 0
---

# MFH

> MFH - OmniData 在 ModelScope 开源的数据集。displayName: MFH license: MIT taskTypes: [] mediaTypes: Image labelTypes: [] tags: attrs: null id: 1257 name: en: Handgesture-recognition zh: 手势识别 attrs: null id: 1258 name: en: Medtech zh: 医疗技术 publisher: Willogy AIOZ…

OmniData/MFH 是 ModelScope 魔搭社区上的2021-01-01、Blood Transfusion Hematology Hospital、https://arxiv.org/pdf/2109.02917.pdf数据集，涉及 publishDate、publisher、paperUrl 领域，存储大小 535 MB，采用 [MIT] 许可。

- **Repository**: OmniData/MFH
- **License**: [MIT]
- **Tasks**: 2021-01-01, Blood Transfusion Hematology Hospital, https://arxiv.org/pdf/2109.02917.pdf, https://github.com/willogy-team/hand-gesture-recognition-smc2021, Image, MFH
- **Domain**: publishDate, publisher, paperUrl, publishUrl, mediaTypes, displayName
- **Storage size**: 535 MB
- **Downloads**: 65
- **Stars**: 0
- **Last updated**: 2024-07-04

Source: https://www.modelscope.cn/datasets/OmniData/MFH

---

displayName: MFH
license:
  - MIT
taskTypes: []
mediaTypes:
  - Image
labelTypes: []
tags:
  - attrs: null
    id: 1257
    name:
      en: Handgesture-recognition
      zh: 手势识别
  - attrs: null
    id: 1258
    name:
      en: Medtech
      zh: 医疗技术
publisher:
  - Willogy
  - AIOZ
  - Blood Transfusion Hematology Hospital
publishDate: '2021-01-01'
publishUrl: https://github.com/willogy-team/hand-gesture-recognition-smc2021
paperUrl: https://arxiv.org/pdf/2109.02917.pdf

---
  ## 简介
  MFH 数据集是一个多视点细粒度的手部卫生数据集。它总共包含 73,1147 个样本，这些样本由 6 个不同位置的 6 个摄像机视图收集。所有样本总共分为 7 个类别。 MFH 数据集与现有数据集的区别在于三个方面：大的类内差异、细微的类间差异以及训练阶段和推理阶段之间的数据分布不匹配。因此，该数据集提供了更现实的基准。
  ## 引文
  ```
@inproceedings{vo2021fine,
  title={Fine-grained Hand Gesture Recognition in Multi-viewpoint Hand Hygiene},
  author={Vo, Huy Q and Do, Tuong and Pham, Vi C and Nguyen, Duy and Duong, An T and Tran, Quang D},
  booktitle={2021 IEEE International Conference on Systems, Man, and Cybernetics (SMC)},
  pages={1443--1448},
  year={2021},
  organization={IEEE}
}
```
  
## Download dataset
:modelscope-code[]{type="git"}