Topics included in this article:
- Camera Positioning and Zooming
- Recent Faces
- Saving and Un-Saving a Face
- Recent Sightings
- Face Details
- Face Alerts
- Fixing an Incorrect Match
- Not a Face Feedback
- Downloading Face Events
- Uploading Faces
- Helpful Links
- Contact Support or Sales
Note: The very first steps in managing your facial recognition will be to make sure you have an enterprise license applied to the camera and then you will need to enable the "Face Recognition" feature.
Camera Positioning and Zooming
Best Practices:
- Mount the camera as close to face level as possible.
- The angle from the face to the camera (from all sides: up, down, right, and left) should be under 30 degrees for optimal setup.
- If the camera is mounted with too harsh of an angle to the face (from any direction), then the feature will not work properly.
- Mount the camera at a distance that will have approximately 100 PPF (pixels per foot) to the person's face at the desired point of capture.
- This article will give you more details on what that PPF translation is to the maximum distance for mounting.
- Note: The maximum distances listed in the linked article above are with maximum resolution and bitrate usage. In most use cases, the distance will need to be smaller than what is listed to reduce bandwidth consumption.
- This article will give you more details on what that PPF translation is to the maximum distance for mounting.
- Mount the camera facing inwards to avoid bad back-lighting.
- Mount the camera in a well-lit region.
- For best results, the whole face should be well-lit with minimal shadows or glaring.
- Make sure to adjust the zoom level of the camera to ensure focus on the intended facial capture region.
- Mount the camera in a region that motivates people passing through to look at the camera for a few seconds.
- A great way to do this is by placing a graphic, sign, etc. near the device to draw the eye. This will allow the device the best vantage point for straight views of the face.
- Note: Many use cases for facial recognition in the field are "uncooperative," meaning the individual is not forced to look at the camera for a length of time in order to perform an action. In cases like these, it is more likely to have suboptimal facial positioning (such as: looking down, off to the side, etc.). Anything that can be done to try to get a straight shot of the face, will enhance the feature's results.
- A great way to do this is by placing a graphic, sign, etc. near the device to draw the eye. This will allow the device the best vantage point for straight views of the face.
Note: If you're using this for unique people counting, we generally recommend mounting above an exit door to capture people as they leave a building.
Recent Faces
Once facial recognition is enabled, you'll notice a "Recent Faces" section on the dashboard. This section will show you faces that any camera with facial recognition enabled has captured. If faces are not showing up, ensure the camera positioning is good, that an enterprise license has been applied, and that the feature has been enabled.
Saving and Un-Saving a Face
Steps to Save a Face
Saving a face means you are building a face profile for that person. To save a face, you simply have to name it.
1. Click on the "Investigations Tab."
2. Navigate to the "Faces" tab.
3. Click on the "Recent Sightings" section.
4. Find the person you want to name.
5. Click on the "Unnamed" link below the face.
6. Enter the desired name for the profile you would like to build.
Note: Once named, the face will appear in the 'Saved Faces' area within the investigations tab and this name can be used for future "Recent Sightings" naming.
Steps to Un-Save a Face
1. Click on the "Investigations Tab."
2. Navigate to the "Faces" tab.
3. Click on the "Recent Sightings" section.
4. Find the capture you want to un-save.
5. Erase the name, leave it blank, and click "Remove Name".
Note: This will now have the recent sighting thumbnail be unnamed and able to be renamed if needed.
You can also remove the name under the "Saved Faces" Tab
1. Click on the name of the profile.
2. Either "Rename" the profile by typing in a new name and clicking "Rename"
-OR-
3. Click "Remove Name" to fully remove the name and fully remove the saved profile.
Note: Once you remove the saved profile, you will need to rebuild it.
Recent Sightings
If you go into Investigations -> Faces, you'll be able to manage the Face Profiles that are saved for your organization.
The recent sightings tab in this section will show you a timeline of thumbnails for all the faces that were recently seen by any camera with a People License enabled. In this section, you can also utilize the search filters to assist with investigations.
If a face was recognized and added to an existing profile, you will see the name attached to the face. If a face is unidentified, in other words not recognized by the system, it will be named 'Unnamed'.
Face Details
- By hovering over a face thumbnail, you can view action options for the captured face (see below). When selecting the time/date stamp at the top, you will be redirected to an "Event Details" pop-up.
- In this pop-up you can view where the event took place, at what time, and what camera tagged the event. Additionally, you can quickly view the related footage by clicking the "View Footage" button:
Face Alerts
Please reference Face Alerts with Labels to learn how to set this up.
Fixing an Incorrect Match
Lighting, positioning, and angle can all affect the initial accuracy of the facial recognition software when making a match. In some cases, an incorrect match may be made. To continue training the model to improve accuracy, it is important to go through the "Recent Sightings" page and correct any misidentifications.
- If a misidentification has occurred, first hover over the thumbnail of the person.
- Select the "Not a Match" button
- The face will now be "Unnamed"
- Click "Unnamed" and begin typing in the correct name. If the name is already a "Saved Face" in the system, it will auto-populate and you can select it.
- Confirm that the image of the saved face is the same person and select "Yes" to merge the profiles. The name will now be correct, and the image will be connected to the correct profile.
Not a Face Feedback
AI models are based on continuous learning. To assist with training the AI, we have exposed a way for you to let us know that the captured face is not actually a face. Doing this improves our AI algorithm and thus offers you better performance. To do this, hover over a face that is not a face and select the "Not a face" button:
Downloading Face Events
You will find a download button at the top right of the "Faces" page in the "Investigations" tab. This button allows you to download face events to a CSV and play around with the data at your leisure.
Uploading Faces
Steps to Upload Faces
1. On the "Faces" page of the "Investigations" tab, you will see an “Upload Faces” button on the top right of the screen.
2. To get the best results, please upload 3-10 photos of each person: driver’s license, agency photo, etc.
3. Ensure that images are sufficiently large in terms of resolution. Faces will be recognized as small as 50 x 50 pixels in image resolutions up to 1920 x 1080. Higher-resolution images require a larger minimum face size. Faces larger than the minimum size provide a more accurate set of facial comparison results.
3. Ensure the file name is exactly how you'd like it to be displayed on the console. Every character in the uploaded file name will directly translate to how it appears in the console.
4. Once John Smith's photos are uploaded, his face will be recognized as John Smith once seen by the camera.
Things to keep in mind while uploading faces
- Use an image with a face that is within the recommended range of angles. The pitch should be less than 30 degrees face down and less than 45 degrees face up. The yaw should be less than 45 degrees in either direction.
- Use an image of a face with both eyes open and visible.
- Use an image of a face that is not obscured or tightly cropped. The image should contain the full head and shoulders of the person. It should not be cropped to the face bounding box.
- Avoid items that block the face, such as headbands and masks.
- Use an image of a face that occupies a large proportion of the image. Images where the face occupies a larger portion of the image are matched with greater accuracy.
- Ensure that images are sufficiently large in terms of resolution. Rhombus can recognize faces as small as 50 x 50 pixels in image resolutions up to 1920 x 1080. Higher-resolution images require a larger minimum face size. Faces larger than the minimum size provide a more accurate set of facial comparison results.
- Use color images.
- Use images with flat lighting on the face, as opposed to varied lighting such as shadows.
Helpful Links
- Features
- Digital Zoom and Pan
- Optics and Object Distances for Analytics
- How to Use the Investigations Tab
- Face Alerts with Labels
- Creating and assigning camera alert policies
Contact Support or Sales
Have more questions? Contact Rhombus Support at +1 (877) 746-6797 option 2 or support@rhombus.com.
Interested in learning more? Contact Rhombus Sales at +1 (877) 746-6797 option 1 or sales@rhombus.com.
Comments
0 comments
Please sign in to leave a comment.