Deep Learning-Based Point Cloud Geometry Coding: RD Control Through Implicit and Explicit Quantization

Guarda, André F. R.; Rodrigues, Nuno M. M.; Pereira, Fernando

http://hdl.handle.net/10400.8/14020

Use this identifier to reference this record.

Name:	Description:	Size:	Format:
Deep learning-based point cloud geometry coding RD control through implicit and explicit quantization.pdf	Deep learning is becoming more and more relevant for multiple multimedia processing tasks, and lately it has raised much interest in the coding arena notably for images and point clouds. While offering near state-of-the-art compression performance, current deep learning-based point cloud coding solutions have a shortcoming since they require training and storing multiple models in order to obtain different rate-distortion trade-offs. This paper proposes a solution that effectively reduces the number of deep learning models that need to be trained and stored by applying explicit quantization to the latent representation, which can be controlled at coding time, to generate varying rate-distortion tradeoffs. The proposed implicit-explicit quantization combination achieves a compression performance that is equivalent or better than the alternative, while significantly reducing the model storage memory requirements.	725.43 KB	Adobe PDF	Download

Send Feedback

Authors

Guarda, André F. R.

Rodrigues, Nuno M. M.

Pereira, Fernando

Abstract(s)

Deep learning is becoming more and more relevant for multiple multimedia processing tasks, and lately it has raised much interest in the coding arena notably for images and point clouds. While offering near state-of-the-art compression performance, current deep learning-based point cloud coding solutions have a shortcoming since they require training and storing multiple models in order to obtain different rate-distortion trade-offs. This paper proposes a solution that effectively reduces the number of deep learning models that need to be trained and stored by applying explicit quantization to the latent representation, which can be controlled at coding time, to generate varying rate-distortion tradeoffs. The proposed implicit-explicit quantization combination achieves a compression performance that is equivalent or better than the alternative, while significantly reducing the model storage memory requirements.

Description

Article number - 9106022; Conference city - London; Conference date - 6 July 2020 - 10 July 2020; Conference code - 162345
EISBN - 978-1-7281-1485-9

Keywords

Point cloud coding deep learning explicit quantization multiple models

URI

http://hdl.handle.net/10400.8/14020

Citation

A. F. R. Guarda, N. M. M. Rodrigues and F. Pereira, "Deep Learning-Based Point Cloud Geometry Coding: RD Control Through Implicit and Explicit Quantization," 2020 IEEE International Conference on Multimedia & Expo Workshops (ICMEW), London, UK, 2020, pp. 1-6, doi: https://doi.org/10.1109/ICMEW46912.2020.9106022.