Quality and Complexity Assessment of Learning-Based Image Compression Solutions

TL;DR

This paper evaluates state-of-the-art learning-based image compression models, comparing their quality and speed to traditional codecs, and finds they offer promising improvements in quality at lower bitrates despite slower speeds.

Contribution

It provides a comprehensive comparison of 8 learning-based models against JPEG2000 and BPG, highlighting their strengths and weaknesses in quality and processing time.

Findings

Learning models outperform JPEG2000 at lower bitrates.

JPEG2000 is faster than learning-based models.

Learning models sometimes surpass BPG in quality metrics.

Abstract

This work presents an analysis of state-of-the-art learning-based image compression techniques. We compare 8 models available in the Tensorflow Compression package in terms of visual quality metrics and processing time, using the KODAK data set. The results are compared with the Better Portable Graphics (BPG) and the JPEG2000 codecs. Results show that JPEG2000 has the lowest execution times compared with the fastest learning-based model, with a speedup of 1.46x in compression and 30x in decompression. However, the learning-based models achieved improvements over JPEG2000 in terms of quality, specially for lower bitrates. Our findings also show that BPG is more efficient in terms of PSNR, but the learning models are better for other quality metrics, and sometimes even faster. The results indicate that learning-based techniques are promising solutions towards a future mainstream compression method.