• Description:

WIDER FACE dataset is a face detection benchmark dataset, of which images are selected from the publicly available WIDER dataset. We choose 32,203 images and label 393,703 faces with a high degree of variability in scale, pose and occlusion as depicted in the sample images. WIDER FACE dataset is organized based on 61 event classes. For each event class, we randomly select 40%/10%/50% data as training, validation and testing sets. We adopt the same evaluation metric employed in the PASCAL VOC dataset. Similar to MALF and Caltech datasets, we do not release bounding box ground truth for the test images. Users are required to submit final prediction files, which we shall proceed to evaluate.

Split Examples
'test' 16,097
'train' 12,880
'validation' 3,226
  • Feature structure:
    'faces': Sequence({
        'bbox': BBoxFeature(shape=(4,), dtype=float32),
        'blur': uint8,
        'expression': bool,
        'illumination': bool,
        'invalid': bool,
        'occlusion': uint8,
        'pose': bool,
    'image': Image(shape=(None, None, 3), dtype=uint8),
    'image/filename': Text(shape=(), dtype=string),
  • Feature documentation:
Feature Class Shape Dtype Description
faces Sequence
faces/bbox BBoxFeature (4,) float32
faces/blur Tensor uint8
faces/expression Tensor bool
faces/illumination Tensor bool
faces/invalid Tensor bool
faces/occlusion Tensor uint8
faces/pose Tensor bool
image Image (None, None, 3) uint8
image/filename Text string


  • Citation:
    Author = {Yang, Shuo and Luo, Ping and Loy, Chen Change and Tang, Xiaoou},
    Booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
    Title = {WIDER FACE: A Face Detection Benchmark},
    Year = {2016} }