AFLW2000-3D Face Detection Dataset

The AFLW2000-3D dataset consists of the first 2000 images of the AFLW database and their 3D information, and is used to evaluate the performance of 3D facial landmark detection models.

These images are annotated with 68-point 3D facial landmarks at the image level. The head poses in this dataset are very diverse, which are usually difficult to detect by CNN-based face detectors.