---
dataset_info:
  features:
  - name: pixel_values
    dtype: image
  - name: label
    dtype: image
  splits:
  - name: train
    num_bytes: 674327851.666
    num_examples: 16706
  - name: validation
    num_bytes: 46935738.0
    num_examples: 1000
  download_size: 797140406
  dataset_size: 721263589.666
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
---

# Dataset Card for Dataset Name

The Active Template Regression (ATR) dataset comprises 18 semantic category labels, including face, sunglasses, hat, scarf, hair, upper clothes, left arm, right arm, belt, pants, left leg, right leg, skirt, left shoe, right shoe, bag, dress, and background. A total of 17,700 images were incorporated into the ATR dataset. 16,700 images were designated for training, and 1,000 for testing.


- **Curated by:** Xiaodan Liang, Si Liu, Xiaohui Shen, Jianchao Yang, Luoqi Liu, Jian Dong, Liang Lin, Shuicheng Yan
- **Shared by:** Xiaodan Liang, Si Liu, Xiaohui Shen, Jianchao Yang, Luoqi Liu, Jian Dong, Liang Lin, Shuicheng Yan
- **License:** MIT

# Dataset Sources

- **Repository:** https://github.com/lemondan/HumanParsing-Dataset
- **Paper:** Deep Human Parsing with Active Template Regression

# Human Parsing Labels

- 0: **background**
- 1: **hat**
- 2: **hair**
- 3: **sunglasses**
- 4: **upperclothes**
- 5: **skirt**
- 6: **pants**
- 7: **dress**
- 8: **belt**
- 9: **leftshoe**
- 10: **rightshoe**
- 11: **face**
- 12: **leftleg**
- 13: **rightleg**
- 14: **leftarm**
- 15: **rightarm**
- 16: **bag**
- 17: **scarf**


# Uses

Semantic segmentation, and more specifically, human body parsing.

# Dataset Card Authors

Christian Kotait

**BibTeX:**

@article{liang2015deep,
  title={Deep human parsing with active template regression},
  author={Liang, Xiaodan and Liu, Si and Shen, Xiaohui and Yang, Jianchao and Liu, Luoqi and Dong, Jian and Lin, Liang and Yan, Shuicheng},
  journal={IEEE transactions on pattern analysis and machine intelligence},
  volume={37},
  number={12},
  pages={2402--2414},
  year={2015},
  publisher={IEEE}
}