Sorting photos and videos by facial recognition

The challenge

Events and media operations can generate large volumes of photos and videos with multiple participants. Manual sorting per person is time-consuming, subject to inconsistencies, and makes it difficult to quickly provide the correct material to each user.

The solution developed

CISP1 developed an application that uses facial detection and recognition to index large media collections and relate occurrences of the same person across different files.

The technology reduces the need for file-by-file inspection and creates an automated search and content organization mechanism, preserving the possibility of review in cases that require additional validation.

How the application works

• Receives the photo and video archive.
• Detects faces and creates suitable representations for comparison.
• Relates occurrences of the same person across different files.
• Filters and organizes the corresponding content.
• Makes the resulting set available for review or delivery.

Value for the operation

• Reduces manual sorting work for large archives.
• Increases the speed of content organization.
• Facilitates locating relevant media for each user.
• Standardizes the classification and retrieval process.

Capabilities applied