---
title: MJU-Waste
canonical_url: "https://www.modelscope.cn/datasets/OmniData/MJU-Waste"
md_url: "https://www.modelscope.cn/datasets/OmniData/MJU-Waste.md"
repository: OmniData/MJU-Waste
last_updated: 2024-07-08
license: "[MIT]"
storage_size: "1.4 GB"
domain:
  - publishDate
  - publisher
  - paperUrl
  - publishUrl
  - mediaTypes
  - displayName
  - taskTypes
tasks:
  - 2020
  - "Fuzhou University"
  - "https://arxiv.org/pdf/2007.04259v1.pdf"
  - "https://github.com/realwecan/mju-waste"
  - Image
  - MJU-Waste
  - "Object Detection"
downloads: 2170
stars: 0
---

# MJU-Waste

> MJU-Waste - OmniData 在 ModelScope 开源的数据集。displayName: MJU-Waste labelTypes: [] license: MIT mediaTypes: Image paperUrl: https://arxiv.org/pdf/2007.04259v1.pdf publishDate: "2020" publishUrl: https://github.com/realwecan/mju-waste publisher: South China…

OmniData/MJU-Waste 是 ModelScope 魔搭社区上的2020、Fuzhou University、https://arxiv.org/pdf/2007.04259v1.pdf数据集，涉及 publishDate、publisher、paperUrl 领域，存储大小 1.4 GB，采用 [MIT] 许可。

- **Repository**: OmniData/MJU-Waste
- **License**: [MIT]
- **Tasks**: 2020, Fuzhou University, https://arxiv.org/pdf/2007.04259v1.pdf, https://github.com/realwecan/mju-waste, Image, MJU-Waste, Object Detection
- **Domain**: publishDate, publisher, paperUrl, publishUrl, mediaTypes, displayName, taskTypes
- **Storage size**: 1.4 GB
- **Downloads**: 2170
- **Stars**: 0
- **Last updated**: 2024-07-08

Source: https://www.modelscope.cn/datasets/OmniData/MJU-Waste

---

displayName: MJU-Waste
labelTypes: []
license:
- MIT
mediaTypes:
- Image
paperUrl: https://arxiv.org/pdf/2007.04259v1.pdf
publishDate: "2020"
publishUrl: https://github.com/realwecan/mju-waste
publisher:
- South China University of Technology
- Minjiang University
- Fuzhou University
tags:
- Waste
taskTypes:
- Semantic Segmentation
- Object Detection

---
# 数据集介绍
  ## 简介
  我们解决了从彩色​​图像和可选的深度图像中定位废弃物体的问题，这是机器人与此类物体交互的关键感知组件。具体来说，我们的方法集成了多个空间粒度级别的强度和深度信息。首先，场景级深度网络产生初始粗分割，在此基础上我们选择一些潜在的对象区域来放大并执行精细分割。上述步骤的结果进一步集成到一个密集连接的条件随机场中，该随机场学习以像素级精度尊重外观、深度和空间亲和力。此外，我们创建了一个新的 RGBD 垃圾对象分割数据集 MJU-Waste，并将其公开以促进该领域的未来研究。我们的方法的有效性在 MJU-Waste 和 Trash Annotation in Context (TACO) 数据集上得到了验证。
  ## 引文
  ```
@article{wang2020multi,
  title={A multi-level approach to waste object segmentation},
  author={Wang, Tao and Cai, Yuanzheng and Liang, Lingyu and Ye, Dongyi},
  journal={Sensors},
  volume={20},
  number={14},
  pages={3816},
  year={2020},
  publisher={Multidisciplinary Digital Publishing Institute}
}
```
  
## Download dataset
:modelscope-code[]{type="git"}