Best Practices for Creating a Face Recognition Model

MediaValet's Face Recognition model relies on a clear, high-quality 200px x 200px image of a face to serve as a baseline for AI recognition. The baseline image should be high-resolution, with the face looking forward and unobstructed.

We suggest creating a dedicated folder for all baseline face images to make sure each important individual within your organization is included.

To create this folder, follow these guides:  
- How to create a new category: https://support.mediavalet.com/hc/en-us/articles/8158048377876

- How to upload assets into your library: https://support.mediavalet.com/hc/en-us/articles/9569583360660

 

Our recommendations:

  1. Name the folder: Face Recognition Images
  2. Upload a high definition, front facing, unobstructed images of each person you want to be recognized by Face Recognition
  3. Upload additional images of the same person into your library to enhance recognition accuracy.

As more images of a face are uploaded to your library, the more accurate the Face Recognition model will be!

 

Although this image has numerous faces, Face Recognition would not be able to identify these faces because each is either obstructed, blocked, or not large enough to meet the 200px by 200px requirement.

AdobeStock_223541783.jpeg

 

This image provides a high quality and clear face for Face Recognition to identify and create a model out of. 

AdobeStock_259097390.PNG

 

 

 

Was this article helpful?
0 out of 1 found this helpful