> For the complete documentation index, see [llms.txt](https://support.hivo.co/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://support.hivo.co/client-profile/client-profile-settings/asset-and-metadata-management/recognition-settings.md).

# Recognition Settings

Configure the facial recognition settings in your account to control the accuracy of who is detected by HIVO's facial recongition technology

## What are Recognition Settings?

The **Recognition Settings** feature allows you to configure how the system **detects, groups, and recognizes faces** across your account. This is useful for tracking important faces, helping you manage the data, usage rights and [**consent**](https://support.hivo.co/consent.md) that relate to the people that appear in your company's photos and videos more effectively.

{% hint style="info" %}
**Setting up " Settings" in HIVO will help:** *Using Recognition Settings, you can fine-tune the HIVO system to focus only on important faces, helping you manage your data efficiently and save time.*
{% endhint %}

## How to use Recognition Settings

* ⚠️ **User Right Required:** User must have the permission: **“Is Client Admin - User can manage the Client Profile”.**
* Navigate to **Recognition Settings: Client Profile** → **Settings** → **Recognition Settings.**

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2F76QnKJOKh5n72bEg6RR6%2FGroup%201151.png?alt=media&amp;token=56316944-3daf-4ced-b34a-724f4a6ad4d8" alt=""><figcaption></figcaption></figure>

{% hint style="warning" %}
**Recognition matching must be enabled by HIVO:** Face matching (recognition vectors) is turned off for new accounts. Until HIVO enables it, **Disable Recognition Vectors** stays on and the page shows a **Verification Required** message. To request it, click **Submit Support Ticket** on that message.

**Only Track Known Faces** and **Minimum Number of Faces to Group** only appear if HIVO has also enabled Anonymous faces for your account. Without it, Only Track Known Faces is always on.
{% endhint %}

### 1. Configure Face Recognition Settings

#### 👉 Option 1: Disable Recognition Vectors

If this setting is **turned ON,** it means that the creation of vector data for new images uploaded to the Library will be disabled.

* **Detection Still Works:** The system will still scan and identify faces in your images, and the bounding boxes (the frames around faces) will still be visible.
* **Manual Assignment is Possible:** You can still manually select these detected appearances and assign them to a HIVO Recognition Profile.
* While this setting is on, **Only Track Known Faces**, **Manual Confirmation Required for Known Faces** and **Minimum Face Similarity** can't be changed.

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2F6Dvjawa4FN6q71jVXyVn%2FGroup%201109.png?alt=media&amp;token=fe54fa93-0c60-4cef-bcc3-2a062c12c025" alt=""><figcaption></figcaption></figure>

#### 🔹 How to completely Disable Recognition for the Account

To effectively shut down vector-based recognition for your entire account, you must complete two steps:

* Enable the **"Disable Recognition Vectors" setting**: This stops the creation of recognition data for all future uploads.
* Enable the “[**Remove Untagged Faces**](https://support.hivo.co/client-profile/client-profile-settings/asset-and-metadata-management/recognition-settings#step-3-additional-actions-remove-untagged-faces)” action: This will wipe any existing vector data from past images that haven't been tagged yet.

After these two steps, your account will no longer have any vector data available to perform automated face comparisons.

#### 👉 Option 2: Only Track Known Faces

* When this option is enabled:
  * Detected faces will **only be assigned to known people** on the **People** page.
  * Unknown faces will: **Not** be grouped into **Anonymous** profiles and **Not** have **Vector Data** stored
* **When to use:**
  * When you only need to track **known individuals**
  * When you want to **enhance privacy** and avoid storing unnecessary face data

{% hint style="info" %}
This option is additional and only appears when a Super User or HIVO Admin simultaneously enables both of the following settings under Additional Features:

* **Allow Recognition Matching Feature and Settings**
* **Allow Anonymous Faces and Settings**
  {% endhint %}

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2F3lMqQZnIifbCagmKtAKm%2FGroup%20939%20(2).png?alt=media&amp;token=6208d679-5aaf-4f76-be41-66bdb8e0ff01" alt=""><figcaption></figcaption></figure>

#### 👉 Option 3: Manual Confirmation Required for Known Faces

When this setting is turned ON, people detected in assets won’t be automatically assigned to existing profiles when a match is found.

* Instead, a suggested match is made for the detection, which may be optionally confirmed or ignored.
* The suggested matches can be found on the [**Approvals - People page**](https://support.hivo.co/approvals/people-approvals.md), or on the Assets themselves as inline suggestions.
* A person in the Asset Library will not come up in searches until the suggested match is confirmed.

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2Fe64okmexImH8lh0Uo1TA%2FGroup%201110.png?alt=media&amp;token=20c53be1-2bb7-4d3b-9857-4a36feed65b6" alt=""><figcaption></figcaption></figure>

### 2. Face Grouping & Detection Settings

#### 2.1. Minimum Number of Faces to Group

* Specify the **minimum number of similar faces** required to save the face as an **Anonymous person.** Once a face appears enough times, it will save as an **Anonymous person**, making it easier to identify the person later from the **People page.**
* **Default value:**
  * Leave blank to use the default value of **4**
  * Suitable for **small to medium** numbers of uploads
* **Recommendations:** If uploading **a large number of images or faces**:
  * Increase the value to **10 or more**
  * Ensures only frequently appearing faces are saved as Anonymous profiles

{% hint style="info" %}
This option is additional and only appears when a Super User or HIVO Admin simultaneously enables both of the following settings under Additional Features:

* **Allow Recognition Matching Feature and Settings**
* **Allow Anonymous Faces and Settings**
  {% endhint %}

#### 2.2. Minimum Face Matching Size (%)

Specify the **minimum size a face needs to be** in order to be detected. Very small faces may not be useful to keep track of.

* **Default value:** Leave blank to use the default value of **0%**
* **Recommendations: 2%** is useful for **ignoring very small faces** in crowd or wide-angle images

#### 2.3. Minimum Face Confidence (%)

Specify the **minimum confidence level** required for a face to be detected.

* **Default value:** Leave blank to use the default value of **99%**
* **Recommendations:**
  * **99% or higher** is recommended
  * Helps prevent **false positives**

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2Fw5XNzFahScb9bgOKin0X%2FGroup%201018.png?alt=media&amp;token=a53690e1-13c6-4a95-b31a-2d7725f087eb" alt=""><figcaption></figcaption></figure>

#### 2.4. Minimum Face Similarity (%)

Specify the **minimum similarity level** required for two faces to be matched as the **same person**. High similarity ensures that similar faces always belong to the same person.

* **Default value:** Leave blank to use the default value of **97%**
* **Recommendations:**
  * **97%** works well in most cases
  * **95%** may work better when uploading **a very small number of faces**, improving match rates

### 3. Configure Minimum Face Sharpness

The **Minimum Face Sharpness** setting lets you control how clear a detected face must be before HIVO includes it in face recognition. This helps filter out blurry faces, such as people in the background, and keeps face recognition results more relevant.

#### Step 1: Open Recognition Settings

* Go to **Client Settings → Recognition Settings**.
* Find **Minimum Face Sharpness**, below **Minimum Face Matching Size** and **Minimum Face Confidence**.

⚠️ **User Right Required:** User must have the permission: **“Is Client Admin - User can manage the Client Profile”.**

#### Step 2: Set Minimum Face Sharpness

* Enter a value between **0 and 100** in the **Minimum Face Sharpness** field, or use the slider to adjust it.
* Review the **Example photo** to see how your selected value affects face detection:
  * **Green boxes:** Faces that meet the minimum sharpness requirement and will be kept.
  * **Red dashed boxes:** Faces that fall below the requirement and will be ignored.
* Click **Save** to apply your changes.

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2Fyuqb7Hewb5N8AdxgFvp5%2FGroup%201168.png?alt=media&amp;token=2f24f7d4-0e2f-4985-85b2-bcef7cbaa4d7" alt=""><figcaption></figcaption></figure>

#### 🔹 Recommended Settings

| Value                    | Behaviour                                                                                   |
| ------------------------ | ------------------------------------------------------------------------------------------- |
| **Blank or 0 (Default)** | Keeps faces regardless of sharpness.                                                        |
| **20–40**                | Filters out noticeably blurry faces, such as people in the background.                      |
| **50 or higher**         | Keeps only sharper, in-focus faces. Useful for event or sports libraries with large crowds. |

#### 🔹 How to disable Minimum Face Sharpness

To stop filtering faces based on sharpness:

* Return to **Client Settings → Recognition Settings**.
* Clear the **Minimum Face Sharpness** field or set the value to **0**. If your organisation uses face recognition to identify people who have declined consent, consider leaving this setting disabled or using a very low value. Blurry faces may otherwise be excluded from recognition (**Consent tracking)**
* Click **Save**.

HIVO will no longer exclude detected faces based on their sharpness scores.

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2FTrzKlFz5HG9UEOk6w5KM%2FGroup%201169.png?alt=media&amp;token=67427bfe-1646-46d2-b9a3-96c8c1db5ead" alt=""><figcaption></figcaption></figure>

#### 🔹 Good to know

* **Images:** Each detected face is evaluated individually. Faces below the selected threshold are excluded from Recognition, People and Approvals.
* **Videos:** A person is kept if their face meets the sharpness requirement in at least one frame.
* **Existing Assets:** Changes only affect new face detections. Previously detected or ignored faces are not automatically updated.
* **Other Recognition Settings:** Minimum Face Sharpness works alongside **Minimum Face Matching Size** and **Minimum Face Confidence**. Faces must meet all three requirements to be kept.

### 4. Create Person Name Field

The **Create Person Name Field** tells HIVO which metadata field holds the person's name on your portraits. **Create Person** reads the name from this field, then tags the single face on the portrait with that person, using an existing People Profile with the same name or creating a new one.

{% hint style="info" %}
**Most accounts don't need to change this.** The default is **Person**, where the IPTC **"Person In Image"** value from your photo files is saved on upload. Only change it if your names are stored in a different field, such as a custom **Staff Name** field.
{% endhint %}

#### 🔹  Step 1: Open Recognition Settings

* Go to **Client Profile → Settings → Recognition Settings**.
* Find **Create Person Name Field**, above **Additional Actions**. It is shown even when **Disable Recognition Vectors** is on.

⚠️ **User Right Required:** User must have the permission: **“Is Client Admin - User can manage the Client Profile”.**

#### 🔹  Step 2: Choose the Name Field

* Open the **Name Field** dropdown. It lists **Person** first, then your account's common and custom metadata fields.
* Select the field that holds **one person's name per portrait**.
* Click **Save** to apply your changes.

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2FwElVVaUWQPvFC9KobMP0%2FGroup%201174.png?alt=media&amp;token=896d2471-7e5a-4cbf-ae37-7c204e9880ec" alt=""><figcaption></figcaption></figure>

#### 🔹 Where this setting is used

| Where you run Create Person                         | Which field it reads                                                                                  |
| --------------------------------------------------- | ----------------------------------------------------------------------------------------------------- |
| **Uploader** (Create Person automation)             | Always this setting.                                                                                  |
| **Library → Actions → Create Person**               | This setting when **Default (Recognition Settings)** is selected. You can pick another field per run. |
| **Automations** (on an Asset, or Library → Actions) | This setting when **Default (Recognition Settings)** is selected. You can pick another field per run. |

#### 🔹 Choosing the right field

* **Pick a field that only holds the person in the photo.** Fields such as **Photographer**, **Author** or **Copyright** hold other people's names: choosing one would create People named after the photographer and tag them on the wrong faces.
* **One name per portrait.** A value such as "Sam Smith, Sam Green" creates one person with that whole text.
* **Two fields with the same label?** Some accounts have two fields whose names only differ in upper and lower case, such as **usage** and **Usage**, and both show the same label. Check which one your assets use before choosing it.

#### 🔹 Good to know

* **Names are tidied up:** extra spaces are removed and matching ignores upper and lower case, so "josh worrell" and "Josh WORRELL" become the same person.
* **Skipped assets:** assets where the field is empty, or the name is longer than **64 characters**, are skipped and counted in the Create Person summary.
* **Existing Assets:** changing the field does not change faces that are already tagged. It only affects the next time Create Person runs.
* **Manual Confirmation:** if **Manual Confirmation Required for Known Faces** is on, Create Person suggests the match instead of tagging it. Review it on the [**Approvals - People page**](https://support.hivo.co/approvals/people-approvals.md).

### 5. Additional Actions - **Remove Untagged Faces**

This tool is designed to purge legacy data, making it especially useful if the system captured anonymous faces **before "Only Track Known Faces" was enabled** or during periods when the setting was toggled.

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2FmBFOv39yJa8U9QwKZvPd%2FGroup%201017.png?alt=media&amp;token=d2016fea-01cd-4d68-8b71-4e1d172298b4" alt=""><figcaption></figcaption></figure>

* Users must explicitly check the box **"I Understand, Remove Untagged"** to confirm the action.
* The system will delete all Vector data associated with untagged faces. Please note that this action is irreversible.

#### 🔹 **Additional Cleanup Options:**

* **Remove Anonymous Faces:** Completely wipes anonymous face profiles from the People page.
* **Remove Bounding Boxes:** Eliminates face detection frames for unidentified individuals when viewing photos or videos.
  * If **UNCHECKED**: The system deletes the recognition data but still retain the bounding boxes on the image (acting as a placeholder for you to tag later).
  * If **CHECKED**: The system will wipe all traces. The bounding boxes surrounding all untagged faces will be completely removed from the Asset.

<figure><img src="https://2446612760-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2F0sjg5GpIPr5HzBTJYqYe%2Fuploads%2FoIxOp0J0hb6TbOp9T1or%2FGroup%20940%20(1).png?alt=media&amp;token=2eac5ef7-586b-492c-94d2-41881206e0e5" alt=""><figcaption></figcaption></figure>

Then, save the changes so this takes effect for future Recognition. You can read more about [**People feature**](https://support.hivo.co/people) **here.**

If you have any questions, don't hesitate to ask, we're always here to help!


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://support.hivo.co/client-profile/client-profile-settings/asset-and-metadata-management/recognition-settings.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
