Studying Very Low Resolution Recognition Using Deep Networks

Abstract

Visual recognition research often assumes a sufficient resolution of the region of interest (ROI). That is usually violated in practice, inspiring us to explore the Very Low Resolution Recognition (VLRR) problem. Typically, the ROI in a VLRR problem can be smaller than $16\times 16$ pixels, and is challenging to be recognized even by human experts. We attempt to solve the VLRR problem using deep learning methods. Taking advantage of techniques primarily in super resolution, domain adaptation and robust regression, we formulate a dedicated deep learning method and demonstrate how these techniques are incorporated step by step. Any extra complexity, when introduced, is fully justified by both analysis and simulation results. The resulting \textit{Robust Partially Coupled Networks} achieves feature enhancement and recognition simultaneously. It allows for both the flexibility to combat the LR-HR domain mismatch, and the robustness to outliers. Finally, the effectiveness of the proposed models is evaluated on three different VLRR tasks, including face identification, digit recognition and font recognition, all of which obtain very impressive performances.

References in corpus (3)

• Deep Networks for Image Super-Resolution with Sparse Prior
• Synthesis-based Robust Low Resolution Face Recognition
• Self-Tuned Deep Super Resolution

Cited by in corpus (17)

• UnitBox: An Advanced Object Detection Network
• An All-in-One Network for Dehazing and Beyond
• Face Recognition in Low Quality Images: A Survey
• Improved Techniques for Learning to Dehaze and Beyond: A Collective Study
• Deep Joint Face Hallucination and Recognition
• When Image Denoising Meets High-Level Vision Tasks: A Deep Learning Approach
• Effects of Image Degradations to CNN-based Image Classification
• Risk Bounds for Robust Deep Learning
• Fully-Coupled Two-Stream Spatiotemporal Networks for Extremely Low Resolution Action Recognition
• Robust Emotion Recognition from Low Quality and Low Bit Rate Video: A Deep Learning Approach
• Robust Nonparametric Regression with Deep Neural Networks
• Delving into Robust Object Detection from Unmanned Aerial Vehicles: A Deep Nuisance Disentanglement Approach
• Student Network Learning via Evolutionary Knowledge Distillation
• Semi-Coupled Two-Stream Fusion ConvNets for Action Recognition at Extremely Low Resolutions
• High Frequency Residual Learning for Multi-Scale Image Classification
• Learning a Mixture of Deep Networks for Single Image Super-Resolution
• MagnifyMe: Aiding Cross Resolution Face Recognition via Identity Aware Synthesis