S2LD: Semi-Supervised Landmark Detection in Low-Resolution Images and Impact on Face Verification

Amit Kumar, Rama Chellappa; Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2020, pp. 758-759

Landmark detection algorithms trained on high resolution images perform poorly on datasets containing low resolution images. This degrades the performance of facial verification, recognition and modeling that rely on accurate detection of landmarks. To the best of our knowledge, there is no dataset consisting of low resolution face images along with their annotated landmarks, making supervised training infeasible. In this paper, we present a semi-supervised approach to predict landmarks on low resolution images by learning them from labeled high resolution images. The objective of this work is to show that predicting landmarks directly on low resolution images is more effective than the current practice of aligning images after rescaling or superresolution. In a two-step process, the proposed approach first learns to generate low resolution images by modeling the distribution of target low resolution images. In the second stage, the model learns to predict landmarks for target low resolution images from generated low resolution images. With extensive experimentation, we study the impact of the various design choices and also show that prediction of landmarks directly in low resolution, improves performance on the critical task of face verification in low resolution images. As a byproduct, the proposed method also achieves state of the art land mark detection results for high resolution images.