Post-Training Quantization Is All You Need to Perform Cross-Platform Learned Image Compression

Dailan He, Ziming Yang, Yan Wang, Yuan Chen, Qi Zhang, Hongwei Qin

Abstract: It has been witnessed that learned image compression has outperformed conventional image coding techniques and tends to be practical in industrial applications. One of the most critical issues preventing it from being practical is the non-deterministic calculation, which makes the probability prediction cross-platform inconsistent and frustrates successful decoding. We propose to solve this problem by introducing well-developed post-training quantization and making the model inference integer-arithmetic-only, which is much simpler than presently existing training and fine-tuning based approaches yet still keeps the superior rate-distortion performance of learned image compression. Based on that, we further improve the discretization of the entropy parameters and extend the deterministic inference to fit Gaussian mixture models. With our proposed methods, the current state-of-the-art image compression models can infer in a cross-platform consistent manner, which makes the further development and practice of learned image compression more promising.

One-sentence Summary: We propose to solve cross-platform inconsistency issue of learned image compression by adopting integer-arithmetic-only post-training quantization, which makes the further development and practice of learned image compression more promising.