The 147th meeting of MPEG took place in Sapporo from 2024-07-15 until 2024-07-19. Find more information here.
MPEG ratifies Eight Edition of ISO Base Media File Format
At the 147th MPEG meeting, MPEG Systems (WG 3) promoted the 8th edition of ISO/IEC 14496-12 ISO base media file format (ISOBMFF) to Final Draft International Standard (FDIS), the final stage of standard development.
The ever-growing expansion of the ISO/IEC 14496-12 ISO base media file format (ISOBMFF) application area has continuously brought new technologies to the standards. During the last couple of years, MPEG Systems (WG 3) has received new technologies on ISOBMFF for more seamless support of ISO/IEC 23009 Dynamic Adaptive Streaming over HTTP (DASH) and ISO/IEC 23000-19 Common Media Application Format (CMAF) leading to the development of the 8th edition of ISO/IEC14496-12.
The new edition of the standard includes new technologies to explicitly indicate the set of tracks representing various versions of the media presentation of a single media for seamless switching and continuous presentation. Such technologies will enable more efficient processing of the ISOBMFF formatted files for DASH manifest or CMAF Fragments.
MPEG ratifies Syntactic Description Language
At the 147th MPEG meeting, MPEG Systems (WG 4) promoted ISO/IEC 14496-34 syntactic description language to Final Draft International Standard (FDIS), the final stage of standard development.
The Syntax Description Language (SDL) has been developed as part of ISO/IEC 14496-1 MPEG-4 Systems to define the syntax of various standards within MPEG-4 and beyond. To enable independent development of and the ability to reference SDL by other standards, MPEG started the development of ISO/IEC 14496-34 syntactic description language, which reached its final milestone at said meeting.
The new standard comes with various improvements, such as clarifications of parsable and non-parsable variable definitions and value coercion, among others. Finally, it includes various examples for a better understanding of the standard.
MPEG reaches First Milestone for Low-Overhead Image File Format
At the 147th MPEG meeting, MPEG Systems (WG 3) promoted Amendment 3 of ISO/IEC 23008-12 low-overhead image file format to Committee Draft Amendment (CDAM), marking the initial stage of standard development.
ISO/IEC 23008-12 image format specification defines generic structures for storing image items and sequences based on ISO/IEC 14496-12 ISO base media file format (ISOBMFF). As it allows the use of various high-performance video compression standards for a single image or a series of images, it has been adopted by the market quickly. However, it was challenging to use it for very small-sized images such as icons or emojis. While the initial design of the standard was versatile and useful for a wide range of applications, the size of headers becomes an overhead for applications with tiny images. Thus, Amendment 3 of ISO/IEC 23008-12 low-overhead image file format aims to address this use case by adding a new compact box for storing metadata instead of the ‘Meta’ box to lower the size of the overhead.
The standard is planned to be completed, i.e., to reach the status of Final Draft Amendment (FDAM), by the end of 2025.
MPEG ratifies Second Edition of Conformance and Reference Software for Compression of Neural Networks
At the 147th MPEG meeting, MPEG Video Coding (WG 4) promoted the 2nd edition of ISO/IEC 15938‑18 conformance and reference software for compression of neural networks to Final Draft International Standard (FDIS), the final stage of standard development.
An increasing number of artificial intelligence applications based on artificial neural networks, such as edge-based multimedia content processing, content-adaptive video post-processing filters, or federated training, need to exchange updates of neural networks (e.g., after training on additional data or fine-tuning to specific content). For this purpose, MPEG developed a second edition of the standard for coding of neural networks for multimedia content description and analysis (NNC, ISO/IEC 15938-17, published in 2024), adding syntax for differential coding of neural network parameters as well as new coding tools. Trained models can be compressed to at least 10-20% for several architectures, even below 3%, of their original size without performance loss. Higher compression rates are possible at moderate performance degradation. In a distributed training scenario, a model update after a training iteration can be represented at 1% or less of the base model size on average without sacrificing the classification performance of the neural network.
In order to facilitate the implementation of the standard, the accompanying standard ISO/IEC 15938-18 has been updated to cover the second edition of ISO/IEC 15938-17. This standard provides a reference software for encoding and decoding NNC bitstreams, as well as a set of conformance guidelines and reference bitstreams for testing of decoder implementations. The software covers the functionalities of both editions of the standard, and can be configured to test different combinations of coding tools specified by the standard.
MPEG reaches First Milestone for Third Edition of Reference Software and Conformance for the Internet of Media Things
At the 147th MPEG meeting, MPEG Coding of 3D Graphics and Haptics (WG 7) reached the Committee Draft (CD) stage of reference software and conformance for the Internet of Media Things (IoMT), the first stage of standard development.
This new edition comprises a comprehensive reference implementation of IoMT functionalities as specified in the MPEG series of Media Things-related standards, outlined across various parts of ISO/IEC 23093. These standards define data formats and APIs for Media Things, enabling robust integration into diverse applications and services. The software implementation features an extensive suite of Media Things, designed to facilitate the collaborative execution of complex artificial intelligence tasks within distributed environments. This standard targets a wide range of applications, such as crowd monitoring with distributed camera networks and forest fire prevention and surveillance. The advanced capabilities of IoMT will significantly enhance the efficiency and effectiveness of such operations, demonstrating the potential of Media Things in real-world scenarios.