<-Use Cases Go to ToC Data Types->
AI Modules (AIM) are interconnected processing units composing an AI Workflow (AW). There are two types of AIM:
- Composite AIMs are AI Modules composed of multiple AI Modules.
- Basic AIMs do not include (or do not expose) AIMs inside.
This Chapter specifies the Composite AIMs if Table 1 .
Table 1 – Composite AIMs of MPAI OSD – Note that V1.1 is currently not approved.
Acronym | AIM Name | Versions |
|
OSD-AVS | Audio-Visual Scene Description | V1.0 | V1.1 |
OSD-VOI | Visual Object Identification | V1.0 |
using a format aligned with the one adopted for the Uses Cases. Other Technical Specifications, such as Context-based Audio Enhancement (MPAI-CAE), Multimodal Conversation (MPAI-MMC), Object and Scene Description (MPAI-OSD), and Portable Avatar Format (MPAI-PAF) specify Composite AIMs that are used in this Technical Specification. Audio Scene Description (OSD-ASD),
This Chapter also specifies the Basic AIMs of Table 2.
Table 2 – Basic AIMs of MPAI OSD – Note that V1.1 is currently not approved.
Acronym | AIM Name | Versions |
|
OSD-AMX | Audio-Visual Scene Multiplexing | V1.0 | |
OSD-AVA | Audio-Visual Alignment | V1.0 | |
OSD-AVE | Audio-Visual Event Description | V1.1 | |
OSD-VCD | Visual Change Detection | V1.1 | |
OSD-SDX | Audio-Visual Scene Demultiplexing | V1.0 | |
OSD-TVS | TV Splitter | V1.1 | |
OSD-VDI | Visual Direction Identification | V1.0 | |
OSD-VII | Visual Instance Identification | V1.0 | |
OSD-VOE | Visual Object Extraction | V1.0 | |
OSD-VOI | Visual Object Identification | V1.0 | |
OSD-VSD | Visual Scene Description | V1.0 | V1.1 |