Occlusion Coherence: Localizing Occluded Faces with a Hierarchical Deformable Part Model

The presence of occluders significantly impacts perfor-mance of systems for object recognition. However, occlu-sion is typically treated as an unstructured source of noise and explicit models for occluders have lagged behind those for object appearance and shape. In this paper we describe a hierarchical deformable part model for face detection and keypoint localization that explicitly models occlusions of parts. The proposed model structure makes it possible to augment positive training data with large numbers of syn-thetically occluded instances. This allows us to easily in-corporate the statistics of occlusion patterns in a discrimi-natively trained model. We test the model on several bench-marks for keypoint localization including challenging sets featuring significant occlusion. We find that the addition of an explicit model of occlusion yields a system that out-performs existing approaches in keypoint localization accu-racy. 1