CompressNet: Generative Compression at Extremely Low Bitrates

Abstract

Compressing images at extremely low bitrates (< 0.1 bpp) has always been a challenging task since the quality of reconstruction significantly reduces due to the strong imposed constraint on the number of bits allocated for the compressed data. With the increasing need to transfer large amounts of images with limited bandwidth, compressing images to very low sizes is a crucial task. However, the existing methods are not effective at extremely low bitrates. To address this need, we propose a novel network called CompressNet which augments a Stacked Autoencoder with a Switch Prediction Network (SAE-SPN). This helps in the reconstruction of visually pleasing images at these low bitrates (< 0.1 bpp). We benchmark the performance of our proposed method on the Cityscapes dataset, evaluating over different metrics at extremely low bitrates to show that our method outperforms the other state-of-the-art. In particular, at a bitrate of 0.07, CompressNet achieves 22% lower Perceptual Loss and 55% lower Frechet Inception Distance (FID) compared to the deep learning SOTA methods.

References in corpus (14)

• Going Deeper with Convolutions
• End-to-end Optimized Image Compression
• The Cityscapes Dataset for Semantic Urban Scene Understanding
• High-Resolution Image Synthesis and Semantic Manipulation with Conditional GANs
• Lossy Image Compression with Compressive Autoencoders
• Stacked What-Where Auto-encoders
• Checkerboard artifact free sub-pixel convolution: A note on sub-pixel convolution, resize convolution and convolution resize
• Variable Rate Image Compression with Recurrent Neural Networks
• Real-Time Adaptive Image Compression
• Rethinking Lossy Compression: The Rate-Distortion-Perception Tradeoff
• Deep Perceptual Compression
• Human Perceptual Evaluations for Image Compression
• An Autoencoder-based Learned Image Compressor: Description of Challenge Proposal by NCTU
• Generative Compression