Abstract:Low-resolution face recognition is challenging due to uncertain input resolutions and the lack of distinguishing details in low-resolution (LR) facial images. Resolution-invariant representations must be learned for optimal performance. Existing methods for this task mainly minimize the distance between the representations of the low-resolution (LR) and corresponding high-resolution (HR) image pairs in a common subspace. However, these works only focus on introducing various distance metrics at the final layer and between HR-LR image pairs. They do not fully utilize the intermediate layers or multi-resolution supervision, yielding only modest performance. In this paper, we propose a novel two-stage multi-scale resolution-adaptive network to learn more robust resolution-invariant representations. In the first stage, the structural patterns and the semantic patterns are distilled from HR images to provide sufficient supervision for LR images. A curriculum learning strategy facilitates the training of HR and LR image matching, smoothly decreasing the resolution of LR images. In the second stage, a multi-resolution contrastive loss is introduced on LR images to enforce intra-class clustering and inter-class separation of the LR representations. By introducing multi-scale supervision and multi-resolution LR representation clustering, our network can produce robust representations despite uncertain input sizes. Experimental results on eight benchmark datasets demonstrate the effectiveness of the proposed method. Code will be released at https://github.com/hhwang98/TMR.

Learning Parallel Canonical Correlations for Scale-Adaptive Low Resolution Face Recognition

Two-Dimensional Canonical Correlation Analysis and Its Application to Face Recognition

Two-Dimensional Discriminant Canonical Correlation Analysis for Face Recognition

Low-Resolution Face Recognition via Coupled

Two-Stage Multi-Scale Resolution-Adaptive Network for Low-Resolution Face Recognition

Piecewise Regularized Canonical Correlation Discrimination for Low-Resolution Face Recognition

Low-Resolution Face Recognition via Coupled Locality Preserving Mappings

Multilinear locality preserving canonical correlation analysis for face recognition

Multiple Scale Canonical Correlation Analysis Networks For Two-View Object Recognition

Super-resolution for Face Recognition Based on Correlated Features and Nonlinear Mappings

Face Hallucination and Recognition Using Kernel Canonical Correlation Analysis

A Novel Super-resolution Approach Based on Supervised Canonical Correlation Analysis

Sparse Representation Based Discriminative Canonical Correlation Analysis for Face Recognition

Two-dimensional canonical correlation analysis and its application in small sample size face recognition

Low-resolution degradation face recognition over long distance based on CCA

Local consistency preserved coupled mappings for low-resolution face recognition

Super-resolution of human face image using canonical correlation analysis

Coupled Discriminant Multi-Manifold Analysis with Application to Low-Resolution Face Recognition.

Randomized Canonical Correlation Discriminant Analysis for Face Recognition.

Coupled cross-regression for low-resolution face recognition

Facial Expression Recognition Based on Modular Two Dimensional Canonical Correlation Analysis