The 148th meeting of MPEG took place in Kemer from 2024-11-04 until 2024-11-08.
MPEG advances AI-based Point Cloud Coding
At its 146th meeting in April 2024, MPEG issued a Call for Proposals (CfP) to explore innovative AI-based Point Cloud Coding technologies, with the goal of enhancing compression techniques for diverse point cloud data. The CfP addressed the full range of point cloud formats, from dense point clouds used in immersive applications to sparse point clouds generated by Light Detection and Ranging (LiDAR) sensors in autonomous driving. With bit depths ranging from 10 to 18 bits, the CfP called for solutions that could meet the precision requirements of these varied use cases.
At the 148th MPEG meeting, MPEG Coding of 3D Graphics and Haptics (WG 7) reviewed six responses to the CfP. The leading proposal distinguished itself with a hybrid coding strategy that integrates end-to-end learning-based geometry coding and traditional attribute coding. This proposal demonstrated exceptional adaptability, capable of efficiently encoding both dense point clouds for immersive experiences and sparse point clouds from LiDAR sensors. With its unified design, the system supports inter-prediction coding using a shared model with intra-coding, applicable across various bitrates without retraining. Furthermore, the proposal offers flexible configurations for both lossy and lossless geometry coding.
Performance assessments highlighted the leading proposal’s effectiveness, with significant bitrate reductions compared to traditional codecs: a 47% reduction for dense, dynamic sequences in immersive applications and a 35% reduction for sparse dynamic sequences in LiDAR data. For combined geometry and attribute coding, it achieved a 40% bitrate reduction across both dense and sparse dynamic sequences, while subjective evaluations confirmed its superior visual quality over baseline codecs. Typically, 0.2 Mbps – 10 Mbps is required to code the geometry of a dynamic PC sequence.
Encouraged by these promising results, MPEG launched a new AI-based Point Cloud Coding standardization project under WG 7. The leading proposal has been chosen as the initial test model, with a working draft and a common test condition document expected shortly after the 148th meeting. WG 7 extends its appreciation to all contributors who participated in the CfP, recognizing the invaluable role their proposals played in shaping the project’s success.
MPEG remains dedicated to advancing AI-driven point cloud coding technologies in upcoming meetings, gathering further insights to improve compression efficiency and quality. This new system is expected to integrate seamlessly with existing AI ecosystems by utilizing AI-based coding methods, delivering globally optimized, high-performance, and scalable 3D data applications.
The standard is expected to be finalized and reach the status of Final Draft International Standard (FDIS) in 2026.
Please, see more details in https://www.mpeg.org/148th-meeting-of-mpeg/