| 1 Function | 2 Reference Model | 3 Input/Output Data |
| 4 SubAIMs | 5 JSON Metadata | 6 Profiles |
| 7 Reference Software | 8 Conformance Texting | 9 Performance Assessment |
1 Functions
Face Identity Recognition (PAF-FIR):
| Receives | Text Object | Text that is related with the Face to be identified. |
| Image Visual Object | Image containing Face to be identified. | |
| Face Time | Time when the face should be identified. | |
| Visual Scene Geometry | Of the scene where the Face is located. | |
| Searches for | Bounding Boxes | That include faces |
| Finds | best match | Between the Faces and those in a database. |
| Produces | Face Identities | Face Instance Identifiers. |
| Bounding Boxes | Bounding Boxes that include faces. |
2 Reference Model
Figure 1 depicts the Reference Model of the Face Identity Recognition AIM.
Figure 1 – Face Identity Recognition AIM
3 Input/Output Data
Table 1 specifies the Input and Output Data of the of the Face Identity Recognition AIM.
Table 1 – I/O Data of Face Identity Recognition AIM
| Input | Description |
| Auxiliary Text | Text with a content related to Face ID. |
| Image Visual Object | An image containing the Face to be identified. |
| Face Time | The Time during which the Face should be identified. |
| Visual Scene Geometry | The Geometry of the Scene where the Face is located. |
| Output | Description |
| Face Identifiers | Associate strings to elements belonging to some levels in a hierarchical classification (taxonomy). |
| Bounding Boxes | The box containing the Face identified. |
4 SubAIMs
No SubAIMs.
5 JSON Metadata
https://schemas.mpai.community/PAF/V1.3/AIMs/FaceIdentityRecognition.json
6 Profiles
No Profiles.
7 Reference Software
7.1 Disclaimers
- This PAF-FIR Reference Software Implementation is released with the BSD-3-Clause licence.
- The purpose of this PAF-FIR Reference Software is to show a working Implementation of PAF-FIR, not to provide a ready-to-use product.
- MPAI disclaims the suitability of the Software for any other purposes and does not guarantee that it is secure.
- Use of this Reference Software may require acceptance of licences from the respective repositories. Users shall verify that they have the right to use any third-party software required by this Reference Software.
7.2 Guide to the PAF-FIR code
Use of this Reference Software for the PAF-FIR AI Module is for developers who are familiar with Python, Docker, RabbitMQ, and downloading models from HuggingFace
PAF-FIR performs face identity recognition with a pretrained FaceNet model; that is, it identifies the faces in a given number of frames per scene by comparison with a dataset of faces.
The PAF-FIR Reference Software is found at the MPAI gitlab site. It contains:
- src: a folder with the Python code implementing the AIM
- Dockerfile: a Docker file containing only the libraries required to build the Docker image and run the container
- requirements.txt: dependencies installed in the Docker image
- README.md: where to find and save weights of face recognition model FaceNet512.
Library: https://github.com/serengil/deepface
7.3 Acknowledgements
This version of the PAF-FIR Reference Software has been developed by the MPAI AI Framework Development Committee (AIF-DC).
8 Conformance Testing
Table 2 provides the Conformance Testing Method for PAF-FIR AIM.
If a schema contains references to other schemas, conformance of data for the primary schema implies that any data referencing a secondary schema shall also validate against the relevant schema, if present and conform with the Qualifier, if present.
Table 2 – Conformance Testing Method for PAF-FIR AIM
| Receives | Text Object | Shall validate against Text Object Schema. |
| Visual Object (Image) | Shall validate against Visual Object Schema. Image Data shall conform with Visual Qualifier. |
|
| Face Time | Shall validate against Time Schema. | |
| Visual Scene Geometry | Shall validate against Visual Scene Geometry Schema. | |
| Produces | Face Instance IDs | Shall validate against Instance ID Schema. |
| Visual Object (Bounding Box) | Shall validate against Bounding Box Schema. Bounding Box Data shall conform with Visual Qualifier. |
9 Performance Assessment
Performance Assessment of an PAF-FIR AIM Implementation shall be performed using a dataset of faces for each face of which the Identity of the face is provided with reference to a Taxonomy.
The Performance Assessment Report of an PAF-FIR AIM Implementation shall include:
- The Identifier of the PAF-FIR AIM.
- The identifier of the face dataset.
- The identifier of the Taxonomy of face identifiers.
- The Performance of the PAF-FIR AIM Implementation expressed by the Accuracy of the Identifiers provided by the output of the PAF-FIR AIM computed on all faces of the dataset referenced in 2 using the Taxonomy referenced in 3.
