MALF Dataset is a face detection database that is used to evaluate face detection models in the wild. The images in this dataset are all from the Internet. It contains 5250 images of 11932 faces. Each image contains a square bounding box, pitch, curl and other postures.

There are about 838 faces in this dataset with a pixel size less than 20*20, accounting for 7%. It also provides information such as gender, whether wearing glasses, whether it is blocked, and whether the expression is exaggerated.

The MALF Dataset was released by the National Pattern Recognition Laboratory in 2014.