Skip to content
Q5/21AI MULTIMEDIA
STANDARDS INDEX

ITU-T SG21 · QUESTION 5

Standards give structure.
Intelligence takes form.

From a concept or a technology to a connected body of standards. Explore nine themes connecting AI-enabled multimedia applications.

Explore nine themes
THE ROLE OF Q5Static diagram

Shared technical foundations for different systems.

Capabilities

Image · video
Speech · text
Multimodal AI

Standardization

Frameworks & needs

Define the scope

Data & formats

Organize information

Tests & metrics

Compare capabilities

Applications

Agents
Digital humans
Industry uses
  1. 01Identify needs
  2. 02Agree approaches
  3. 03Support collaboration
Q5 develops standards for AI-enabled multimedia: identifying shared needs, agreeing on frameworks and data requirements, and establishing evaluation methods to support collaboration.Q5 scope ↗

Ask about standards, terminology and applications. Follow the source references in each answer.

09Themes
100Catalog records
27Planned media slots

51 Recommendation families43 Project records6 Candidate records

01

Nine themes, one knowledge map

Browse themes
017 catalog records

Foundation Models Evaluation & Assessment

Explore how foundation models are tested, compared and assessed, from general capabilities to multimodal tasks.

Foundation model / Benchmarking
0225 catalog records

Foundation Model Platforms & Engineering Capabilities

Explore the engineering capabilities of foundation model platforms across data preparation, training, deployment and operation.

Model platform / Data annotation
037 catalog records

AI Agents

Explore agents, their collaboration and applications through perception, memory, planning and tool execution.

Agents / Memory
045 catalog records

Hardware & Computing Power Benchmark

Understand computing benchmarks and energy-efficiency measurements through processors, clusters and interconnect architectures.

Computing power benchmark / DNN processor
053 catalog records

Embodied AI

Connect model capabilities with perception and action in the physical world, and explore system frameworks, data and testing.

Embodied AI / Teleoperation
0612 catalog records

Generative AI & Multimedia Processing

Explore text, audio and multimodal generation applications, alongside retrieval augmentation and media processing systems.

AIGC / RAG
0710 catalog records

Digital Human Systems

Understand digital human systems and interactions, from appearance, voice and motion to platforms and interfaces.

Digital humans / Rendering
0813 catalog records

Federated Learning & Distributed AI

Explore model collaboration across participants and the architecture and evaluation of device-edge-cloud systems.

Federated learning / Participants
0918 catalog records

Vertical Industry & Knowledge Engineering

Explore industry scenarios through domain data and knowledge organization, connecting practical problems to standardization.

Knowledge engineering / Domain-specific dataset
02

Explore the catalog

View all

The catalog distinguishes Recommendations, projects and candidates, with unknown status and full-text gaps kept explicit.

03

See the ideas behind the technology

View media plan
Diagram · media to be added
image

Evaluation & assessment · Concept diagram

Dimensions and workflow of model evaluation

Animation · media to be added
animation

Evaluation & assessment · Process animation

How inputs, tasks, metrics and evaluation results connect

Video · media to be added
video

Evaluation & assessment · Introduction & case video

Model evaluation methods and practical testing scenarios

Questions answered from ITU sources