Skip to content
English
Create Support Ticket
  • There are no suggestions because the search field is empty.

What is the Face Recognition Module?

Powerful face recognition with pre-trained and custom models, plus face indexing.

Module Description

Face Recognition detects and identifies the faces. Using the pretrained celebrity dataset, this module covers more than 20.000 personalities, including the world's most famous people and a vast majority of German politicians and athletes.

Customized Facerecognition
To use a custom face recognition model, you need to access the training function within the Deep Model Customizer.

How does it work?

  1. Select the Media File: Choose the media file you want to analyze.
  2. Activate the Face Recognition Module: In the left column, select the "Face Recognition" module.
  3. Define the Model & Parameters: Choose the model for analysis from the available options, set the parameters, and click the yellow "Add Module" button.
  4. Start the Analysis: You can either add more modules or begin the analysis immediately by clicking "Start Analysis"

What Parameters are available?

  • Model (Dropdown): Select from our pretrained celebrity model or your custom-trained models for face recognition. The capabilities are limited to the chosen model

Custom Models:
To use custom models, they must be created and trained in the Deep Model Customizer, after which they will appear in the dropdown.

  • Distance (0-2): Adjusts the strictness of face recognition. Higher values result in a broader match range.

  • Index Unknown Identities (Checkbox): Enables the Deep Indexer to assign unique IDs to unidentified faces across all assets. Renaming an indexed face will automatically update previously indexed images and future analyses.

Face Index:
The Face Index offers the easiest way to manage unknown faces. Each face is automatically assigned a unique ID, allowing you to rename it instantly. Normally, in the Deep Model Customizer, you would need to upload training material for each person. However, with the Face Index, every face becomes recognizable right away without the need for extra training data.

  • Cluster Unknown Identities (Checkbox): Groups unrecognized faces as "unknown" without assigning IDs. All unrecognized individuals will remain unnamed.

  • Face Detection Threshold (0-1): A confidence level for detecting faces. A lower value provides more accurate results, a higher value may lead to more false positives.

  • Face Detection Scale (128-2048): Defines the minimum face size for recognition. Increasing the value helps detect smaller faces.

  • Enable Top-K Prediction (Checkbox): Displays the closest (k) number of matches for a face, allowing manual verification of the results.

  • Top-K (0-10): Defines the number of closest matches to display.

Celebrity comparison:
Ever wondered which celebrity looks most like you?
Activate the Top K Prediction, upload an image and try it out.

Displaying the Results:

Timeline:

The timeline, located below the player, displays the entire video runtime and the results from each module as gray bars.
  • By clicking on any of the grey result bars, you will see details such as:
    • Timecode (TC)
    • Exact frame numbers
    • Runtime/Duration
  • Additional information may also be displayed depending on the module. Clicking on a result moves the playhead to the beginning of that result.
  • These results are identical to those provided by the API, but in a more user-friendly, graphical format. If there are multiple results, use your mouse wheel to scroll through the timeline.

Search Field:

Located in the top bar, the search field includes filter settings for refining your results.
  • Name field: Enter a name to view results that either match or don't match the entered name.
  • Sorting: Results can be sorted alphabetically, by similarity, or by duration. You can toggle between ascending and descending order.
  • Similarity: The similarity slider filters results based on confidence levels, displaying only results above a certain similarity threshold.

After adjusting filters, click "Apply" to apply them. Active filters appear in a black box beneath the search field and can be cleared by clicking the X symbol.

Module Section

On the right side of the player, you’ll see a section with detailed results for each module used in the analysis. Clicking on the module name opens a dropdown with specific parameters, useful for troubleshooting or viewing metadata.

Result Cards

Results are displayed as cards in chronological order. Each card provides key information, such as:


    • Name of the result: Whether the person is identified or marked as "Unknown."

      • If the Deep Indexer was activated, you may see "Unknown with ID number" for clustered identities.
      • If you’ve enabled Cluster Unknown Identities, a single result will group all unknowns together.

    • Knowledge Graph Link:

      If the person is listed in the Deep Explorer Knowledge Graph, you’ll see a Knowledge Graph icon (a ball of dots and lines). Clicking it opens a pop-up showing all related information.

    • Rename:

      Click on the three dots (...) and choose "Rename" to relabel the result. This change will apply to both current and past recognized assets of the person.

    • Go to Training Source Image:

      Click on the three dots (...) and choose "Go to Training Source Image" for accessing the Deep Model Customizer. This option appears when a recognized person is part of a custom dataset. Clicking it takes you to the Dataset section, where you can see the training image responsible for the positive result. This is especially useful for troubleshooting false positives.

Finding the correct training source image:
This helps you in case you are troubleshooting a false positive match of a person and want to locate the training image responsible for the missmatch.

    • Show in Index Collection:

      Click on the three dots (...) and choose "Show in Index Collection:" for accessing the Deep Indexer. This takes you to the Index section of the Deep Indexer, where you can view the indexed person’s ID Card and Unique ID.

The Deep Indexer:
Using the Deep Indexer you can find all other assets where this person was recognized in the past, you can relabel the person there and also deleting the Indexed Person from all assets.
Meaning that the person will not remain in the Index but the media will still be there.

Troubleshooting

To investigate issues with your Face Recognition results, follow these steps:

  1. Use the Module Section:

    • In the module section on the right-hand side of the Result Viewer, you can access detailed parameters for each module. These parameters can provide additional information, which can help with troubleshooting by allowing you to verify settings or fine-tune the module performance.

  2. Go to Training Source Image:

    • If you're dealing with false positive results (where the AI has incorrectly identified someone or something), use the "Go to Training Source Image" option. This will take you to the Dataset section, where you can view the training image responsible for the false match and investigate potential issues with the dataset or labeling.