Robust-FEC-CNN Results of Face Datasets

This repository provides facial landmark detection results of several face datasets by the technique of Robust-FEC-CNN (Won 2nd of CVPR 2017 Faces "In-The-Wild" Workshop-Challenge). These landmarks can be used for aligning faces of these datasets (use align.py).

Format

• bbox.txt

...
image x_min y_min x_max y_max
...

• landmark.txt

...
image x_1 y_1 ... x_i y_i ... x_68 y_68
...

Landmarks File of Face Datasets

Attribute

• CelebA [Google Drive] [Baidu Netdisk]

  

Age

• CACD [Google Drive] [Baidu Netdisk]

  

• IMDB-WIKI [Google Drive] [Baidu Netdisk]

  

• FG-NET [Google Drive] [Baidu Netdisk]

  

• MORPH [Google Drive] [Baidu Netdisk]

  

Identity

• CASIA-WebFace [Google Drive] [Baidu Netdisk]

  

• LFW [Google Drive] [Baidu Netdisk]

  

Citation

If you find the results of Robust-FEC-CNN useful in your research work, please consider citing:

@InProceedings{he2017robust,
    author = {He, Zhenliang and Zhang, Jie and Kan, Meina and Shan, Shiguang and Chen, Xilin},
    title = {Robust FEC-CNN: A High Accuracy Facial Landmark Detection System},
    booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
    year = {2017}
}

@InProceedings{he2017fully,
    author={He, Zhenliang and Kan, Meina and Zhang, Jie and Chen, Xilin and Shan, Shiguang},
    title={A Fully End-to-end Cascaded CNN for Facial Landmark Detection},
    booktitle={The IEEE International Conference on Automatic Face \& Gesture Recognition (FG)},
    year={2017}
}