The WIDER FACE dataset is a face detection benchmark dataset, where images are selected from the open source WIDER dataset. 32k images are chosen and 394k faces annotated with a high degree of variability in scale, pose and occlusion in the sample images. WIDER FACE dataset is organized based on 61 event classes, which are randomly split in a train/val/testing split of 40/10/50.
WIDER Face Dataset
DOWNLOAD WIDER Face (Medium)