Facial Landmark Detection with Tweaked Convolutional Neural Networks

TL;DR

This paper introduces a novel CNN architecture called TCNN for facial landmark detection, which improves accuracy by adaptively tweaking processing based on facial alignment without complex multi-part models.

Contribution

The paper proposes the Tweaked CNN (TCNN), a new CNN design that enhances facial landmark detection by mid-network feature adjustment based on facial alignment.

Findings

TCNN surpasses previous state-of-the-art on benchmark datasets.

The approach improves robustness and accuracy without multi-scale models.

Features from specialized layers effectively capture landmark locations.

Abstract

We present a novel convolutional neural network (CNN) design for facial landmark coordinate regression. We examine the intermediate features of a standard CNN trained for landmark detection and show that features extracted from later, more specialized layers capture rough landmark locations. This provides a natural means of applying differential treatment midway through the network, tweaking processing based on facial alignment. The resulting Tweaked CNN model (TCNN) harnesses the robustness of CNNs for landmark detection, in an appearance-sensitive manner without training multi-part or multi-scale models. Our results on standard face landmark detection and face verification benchmarks show TCNN to surpasses previously published performances by wide margins.