Frequency modulation improves facial landmark detection across datasets
FreqFLD: Towards All-in-One Facial Landmark Detection via Frequency Modulation
Computer Vision and Pattern Recognition
Summary
Facial landmark detection means finding key points on faces, like eyes and mouth, which is important for many apps but can be hard when faces change a lot. Most current methods focus on looking at the face as a whole picture, missing important details about different frequencies or patterns in the image. The authors created a new method called FreqFLD that looks at both the big picture and small details by separating low and high frequency parts of the face data. This approach helps the system better handle different types of faces and conditions, making facial landmark detection more reliable across various datasets.
What this means in practice
- •For mobile app developers: Improve face-based feature recognition in apps by using frequency-modulated landmark detection to work robustly across different users and lighting conditions.
- •For security system integrators: Enhance facial recognition systems in security devices by adapting facial landmark detection to diverse environments and subjects using frequency-aware models.
Authors
Shun Ren, Kaijie Jin, Shengkai Hu, Beihang Song, Hang Sun, Wenwen Min, Youfa Liu, Jun Wan
Abstract
Recent progress in deep learning has significantly advanced facial landmark detection. However, most existing methods process features in a spatial-domain manner under a dataset-specific training paradigm, which overlooks the fact that facial landmark detection is inherently geometry-driven and sensitive to frequency variations, thereby limiting cross-dataset generalization under complex scenarios and hindering the development of a facial landmark detection model. To address this issue, we propose \textbf{FreqFLD}, a \textbf{freq}uency-modulated framework towards All-in-One \textbf{f}acial \textbf{l}andmark \textbf{d}etection. Specifically, FreqFLD introduces a Frequency Modulation Module (FreqMoM) to explicitly induce the frequency prior by decoupling and modulating low- and high-frequency components, which is then injected into subsequent feature modeling to enable balanced modeling of global facial structure and local landmark details. Furthermore, FreqFLD employs a Frequency-Modulated Mixture-of-Experts (FreqMoE), with expert selection adaptively conditioned on frequency-modulated priors, enabling flexible modeling of heterogeneous facial landmark patterns under diverse and challenging scenarios. To regularize frequency-consistent modeling under the All-in-One paradigm, we further introduce a Frequency-Consistent Routing (FreqCR) loss, which constrains the routing and assignment of frequency-aware experts to promote balanced expert utilization across diverse facial scenarios, thereby enabling stable expert specialization and achieving robust facial landmark detection. Extensive experiments demonstrate that the proposed FreqFLD achieves comparable performance on popular datasets. The code is available at: https://github.com/jkj1059657014/FreqFLD.