Loading [MathJax]/extensions/tex2jax.js

Learned Compression in Adaptive Point Cloud Streaming: Opportunities, Challenges and Limitations

Research output: Chapter in book/report/conference proceedingConference contributionResearchpeer review

Authors

  • Michael Rudolph
  • Amr Rizk

Details

Original languageEnglish
Title of host publicationProceedings of the 16th ACM Multimedia Systems Conference
Subtitle of host publicationMMSys 2025
Pages328-334
Number of pages7
ISBN (electronic)9798400714672
Publication statusPublished - 31 Mar 2025
Event16th ACM Multimedia Systems Conference, MMSys 2025 - Stellenbosch, South Africa
Duration: 31 Mar 20254 Apr 2025

Abstract

Learned point cloud compression methods have achieved rate-distortion performance, which is comparable to or higher than conventional approaches. However, this often comes at the cost of high hardware requirements and thus low throughput during encoding and decoding. In this paper, we present an adaptive bitrate point cloud streaming system utilizing learned compression. While other learned compression techniques require to split geometry and attributes, resulting in high encoding latency, we deploy a unified model to handle both modalities together, which drastically reduces the coding complexity. We explore the capabilities of the learned encoder to derive multiple quality representations with only re-running a fraction of the encoding steps, making it a suitable fit for adaptive bitrate streaming. Furthermore, we ablate the encoding latency of each component in the encoder and decoder stack, identifying bottlenecks in the process.

Keywords

    adaptive bitrate streaming, learned point cloud compression, point cloud streaming

ASJC Scopus subject areas

Cite this

Learned Compression in Adaptive Point Cloud Streaming: Opportunities, Challenges and Limitations. / Rudolph, Michael; Rizk, Amr.
Proceedings of the 16th ACM Multimedia Systems Conference: MMSys 2025. 2025. p. 328-334.

Research output: Chapter in book/report/conference proceedingConference contributionResearchpeer review

Rudolph, M & Rizk, A 2025, Learned Compression in Adaptive Point Cloud Streaming: Opportunities, Challenges and Limitations. in Proceedings of the 16th ACM Multimedia Systems Conference: MMSys 2025. pp. 328-334, 16th ACM Multimedia Systems Conference, MMSys 2025, Stellenbosch, South Africa, 31 Mar 2025. https://doi.org/10.1145/3712676.3719266
Rudolph, M., & Rizk, A. (2025). Learned Compression in Adaptive Point Cloud Streaming: Opportunities, Challenges and Limitations. In Proceedings of the 16th ACM Multimedia Systems Conference: MMSys 2025 (pp. 328-334) https://doi.org/10.1145/3712676.3719266
Rudolph M, Rizk A. Learned Compression in Adaptive Point Cloud Streaming: Opportunities, Challenges and Limitations. In Proceedings of the 16th ACM Multimedia Systems Conference: MMSys 2025. 2025. p. 328-334 doi: 10.1145/3712676.3719266
Rudolph, Michael ; Rizk, Amr. / Learned Compression in Adaptive Point Cloud Streaming : Opportunities, Challenges and Limitations. Proceedings of the 16th ACM Multimedia Systems Conference: MMSys 2025. 2025. pp. 328-334
Download
@inproceedings{1e053b7ddb6149bb91d2882a19289ddc,
title = "Learned Compression in Adaptive Point Cloud Streaming: Opportunities, Challenges and Limitations",
abstract = "Learned point cloud compression methods have achieved rate-distortion performance, which is comparable to or higher than conventional approaches. However, this often comes at the cost of high hardware requirements and thus low throughput during encoding and decoding. In this paper, we present an adaptive bitrate point cloud streaming system utilizing learned compression. While other learned compression techniques require to split geometry and attributes, resulting in high encoding latency, we deploy a unified model to handle both modalities together, which drastically reduces the coding complexity. We explore the capabilities of the learned encoder to derive multiple quality representations with only re-running a fraction of the encoding steps, making it a suitable fit for adaptive bitrate streaming. Furthermore, we ablate the encoding latency of each component in the encoder and decoder stack, identifying bottlenecks in the process.",
keywords = "adaptive bitrate streaming, learned point cloud compression, point cloud streaming",
author = "Michael Rudolph and Amr Rizk",
note = "Publisher Copyright: {\textcopyright} 2025 Copyright held by the owner/author(s).; 16th ACM Multimedia Systems Conference, MMSys 2025, MMSys 2025 ; Conference date: 31-03-2025 Through 04-04-2025",
year = "2025",
month = mar,
day = "31",
doi = "10.1145/3712676.3719266",
language = "English",
pages = "328--334",
booktitle = "Proceedings of the 16th ACM Multimedia Systems Conference",

}

Download

TY - GEN

T1 - Learned Compression in Adaptive Point Cloud Streaming

T2 - 16th ACM Multimedia Systems Conference, MMSys 2025

AU - Rudolph, Michael

AU - Rizk, Amr

N1 - Publisher Copyright: © 2025 Copyright held by the owner/author(s).

PY - 2025/3/31

Y1 - 2025/3/31

N2 - Learned point cloud compression methods have achieved rate-distortion performance, which is comparable to or higher than conventional approaches. However, this often comes at the cost of high hardware requirements and thus low throughput during encoding and decoding. In this paper, we present an adaptive bitrate point cloud streaming system utilizing learned compression. While other learned compression techniques require to split geometry and attributes, resulting in high encoding latency, we deploy a unified model to handle both modalities together, which drastically reduces the coding complexity. We explore the capabilities of the learned encoder to derive multiple quality representations with only re-running a fraction of the encoding steps, making it a suitable fit for adaptive bitrate streaming. Furthermore, we ablate the encoding latency of each component in the encoder and decoder stack, identifying bottlenecks in the process.

AB - Learned point cloud compression methods have achieved rate-distortion performance, which is comparable to or higher than conventional approaches. However, this often comes at the cost of high hardware requirements and thus low throughput during encoding and decoding. In this paper, we present an adaptive bitrate point cloud streaming system utilizing learned compression. While other learned compression techniques require to split geometry and attributes, resulting in high encoding latency, we deploy a unified model to handle both modalities together, which drastically reduces the coding complexity. We explore the capabilities of the learned encoder to derive multiple quality representations with only re-running a fraction of the encoding steps, making it a suitable fit for adaptive bitrate streaming. Furthermore, we ablate the encoding latency of each component in the encoder and decoder stack, identifying bottlenecks in the process.

KW - adaptive bitrate streaming

KW - learned point cloud compression

KW - point cloud streaming

UR - http://www.scopus.com/inward/record.url?scp=105005026528&partnerID=8YFLogxK

U2 - 10.1145/3712676.3719266

DO - 10.1145/3712676.3719266

M3 - Conference contribution

AN - SCOPUS:105005026528

SP - 328

EP - 334

BT - Proceedings of the 16th ACM Multimedia Systems Conference

Y2 - 31 March 2025 through 4 April 2025

ER -