Facial Landmarks Localization using Cascaded Neural Networks
This work addresses facial landmark localization for applications like face recognition and expression analysis, representing an incremental improvement over existing methods.
The paper tackled the problem of accurately localizing facial landmarks for face analysis tasks by proposing a deep learning architecture with cascaded subnetworks, achieving favorable comparison with state-of-the-art schemes, particularly in challenging conditions.
The accurate localization of facial landmarks is at the core of face analysis tasks, such as face recognition and facial expression analysis, to name a few. In this work, we propose a novel localization approach based on a deep learning architecture that utilizes cascaded subnetworks with convolutional neural network units. The cascaded units of the first subnetwork estimate heatmap-based encodings of the landmarks locations, while the cascaded units of the second subnetwork receive as input the output of the corresponding heatmap estimation units, and refine them through regression. The proposed scheme is experimentally shown to compare favorably with contemporary state-of-the-art schemes, especially when applied to images depicting challenging localization conditions.