Skip to yearly menu bar Skip to main content


Poster

Learning to Inpaint for Image Compression

Mohammad Haris Baig · Vladlen Koltun · Lorenzo Torresani

Pacific Ballroom #122

Keywords: [ Deep Learning ] [ Deep Autoencoders ]


Abstract:

We study the design of deep architectures for lossy image compression. We present two architectural recipes in the context of multi-stage progressive encoders and empirically demonstrate their importance on compression performance. Specifically, we show that: 1) predicting the original image data from residuals in a multi-stage progressive architecture facilitates learning and leads to improved performance at approximating the original content and 2) learning to inpaint (from neighboring image pixels) before performing compression reduces the amount of information that must be stored to achieve a high-quality approximation. Incorporating these design choices in a baseline progressive encoder yields an average reduction of over 60% in file size with similar quality compared to the original residual encoder.

Live content is unavailable. Log in and register to view live content