Overview
ISO/IEC TR 23888-1:2026 is an internationally recognized technical report published by ISO/IEC under the title, Information technology - Artificial intelligence for multimedia - Part 1: Vision and scenarios. This standard provides a foundational overview of the role of artificial intelligence (AI), including neural networks (NN), in multimedia coding and processing. It explains current perspectives on utilizing AI for multimedia applications, identifies key technical challenges, and outlines scenarios and gaps that future standardization efforts should address. The document is an essential reference for organizations and professionals engaged in the integration of AI within multimedia systems, seeking interoperability, efficiency, and reproducibility in evolving digital media environments.
Key Topics
- AI and Multimedia Interplay: Discusses how AI technologies, particularly neural networks, are used both as tools for multimedia coding (such as AI-based video compression) and as consumers of multimedia data (such as video coding optimized for machine learning tasks).
- Technical Challenges: Identifies challenges including reproducibility of AI model results, achieving interoperability across diverse hardware and software platforms, and managing computational complexity and efficiency.
- Coding Activities: Explores AI-driven activities such as video coding, 3D graphics coding, model compression, and feature coding to facilitate robust, scalable multimedia processing.
- Performance Metrics: Highlights the need for new or adapted metrics (beyond traditional video compression measures) to accurately assess the impact and utility of AI components.
- Standardization Gaps: Pinpoints technical and architectural areas lacking current standards, specifically for neural network-based media processing and interoperable model formats.
Applications
The standard demonstrates practical applications for AI in multimedia by defining and describing typical scenarios:
- AI-Based Video Coding: Neural networks are increasingly used to optimize video compression, enabling reduced bit rates without sacrificing quality. Hybrid codecs, which integrate traditional and AI-based tools, deliver benefits to content streaming, broadcasting, and video conferencing solutions.
- 3D Graphics and Point Cloud Coding: AI enhances the representation and compression of 3D data, facilitating realistic virtual, augmented, and mixed reality experiences in fields like entertainment, healthcare, and industrial design.
- Model Compression and Deployment: Efficient exchange and deployment of compressed AI models enables real-time edge processing in smart cameras, mobile devices, and IoT deployments.
- Downstream AI Tasks: Tailoring video coding for AI-based analysis, such as object detection, allows more effective application of machine learning in surveillance, automotive, and smart city scenarios.
- Feature Coding for AI Models: The ability to encode and transmit features optimally between distributed AI systems supports scalable solutions for connected devices and distributed intelligence.
Related Standards
ISO/IEC TR 23888-1:2026 is closely linked with several relevant standards in the multimedia and artificial intelligence domain:
- ISO/IEC 15938-15 & ISO/IEC 15938-17: Address AI-driven compact video descriptors and neural network compression, facilitating efficient exchange and deployment of trained models.
- ISO/IEC 23002-7: Specifies supplemental enhancement information messages for coded video bitstreams, crucial for the delivery and signaling of neural network-based enhancements.
- ISO/IEC 22989: Provides primary AI concepts, definitions, and terminology, ensuring consistent understanding across AI-related standards.
- MPEG-AI (ISO/IEC 23888 Series): Serves as the parent series, with further parts expanding on practical use cases and technical solutions for AI integration in multimedia.
Conclusion
ISO/IEC TR 23888-1:2026 is an authoritative reference for integrating artificial intelligence in multimedia coding and processing. It addresses practical scenarios, promotes industry interoperability, and points to the future needs of AI-powered multimedia systems. Key stakeholders ranging from multimedia codec developers to content distributors and device manufacturers will benefit from understanding and applying this standard, paving the way for further innovation and standardization in AI-driven multimedia technologies.