Face
Welcome to the Face Forum
Categories
API – Any ideas or feedback pertaining to features or enhancements to Face API.
Documentation – Any ideas or suggestions for the API Reference or Documentation.
Language Support – Submit a request to have a particular language supported.
Samples & SDK Request – Let us know if you would like to see a Code sample or SDK provided.
Attention!
We have moved our Customer Feedback & Ideas for Azure Cognitive Services portal to the Azure Feedback Forum.
-
Can we buy a license for an FaceAPI SDK that works offline?
In some case, we would prefer to use the face API offline especially in the financial and Pension services industry. We would like to run face API on out customer profile images. We might not be allowed to upload these to an external server.
61 votes -
Support for more than 1,000 person objects in Face API
Need a way to go beyond 1,000 person objects with the face api when doing identification tasks.
39 votesWhile personGroups are still limited to 1,000 people, you are now able to add up to 1,000 personGroups to the standard tier, for a total of 1 million person objects.
personGroups are now limited to 10,000 people.
-
Get also age and gender with the Emotion API
If we want to get the age or gender along with the emotion of a person, we need to call different APIs. I think it is a common scenario, so having all this information with a single Emotion API could be a great improvement.
28 votes -
A way to determine if the image is a photo or a real person
A confidence value returned in the json to indicate whether the face detected/recognised is not a spoof attempt. eg a person is not holding up a photo to the camera
15 votes -
Add liveliness detection
Please add liveliness detection (or blink checking / moving face) to FaceAPI so that i can be used for authentication.
I understand uber use face api to verify drivers, however in personal tests ive seen you can easily get around the detection by taking a photo of a photo.
15 votes -
Increase number of candidates for Identify method
Hi,
Currently in method Identify the paramenter maxNumOfCandidatesReturned can be set between 1 and 5.We would like to increase maxNumOfCandidatesReturned to 1000 and get all persons in Identify
15 votes -
Identify if person is looking at the camera
Hi, I'd like to know if it is possible to add a feature that tells is the recognized face is looking into the camera or very near to it. This would improve the stuff we do here greatly.
thanks!14 votes -
Face API: Custom confidence threshold for Find Similar / Identify
For the usecase of "find the most similar face from a group or face list, even if it's not the same person", it would be nice to be able to provide custom thresholds for Find Similar and Identify.
There seems to be a fairly high internal threshold by default, the idea is to be able to override that.
Also please include default thresholds in the docs (if any).14 votesHappy to announce this was released to the service as part of our July 2016 update and we recently updated the C#, Python, iOS, and Android packages on GitHub to support these changes as well.
Finding Similar Face API: Added optional mode field enabling selection of two working modes, default matchPerson works the same as before, and new mode matchFace removes the same person filtering. If mode field is not specified, the behavior is the same as the past release.
Face Identification API: Optional user-specified confidenceThreshold is enabled for user to define the confidence threshold of whether one face belong to a person object.
Please take a look and provide feedback or requests for additional features.
Thanks!
-
Make confidence level available for age and for gender
Actually confidence level is only available when identification is performed, it would be cool to also have confidence level for age and gender.
12 votes -
Hair Analysis
Hair analysis. hair color(blonde,...), hair type(long, nohair, short hair,..)
12 votes -
/verify should also take persistedFaceIds
Currently /verify only takes faceIds, it could also take persistedFaceIds from personobjects which would improve ease of use
11 votes -
Face detect automatic orientation correction
Faces can be on their side or upside down depending upon EXIF orientation or general composition of the photo. For example, a landscape photo of baby lying on its back will have the baby's face on the side. Right now, the face api does not correctly detect the face when it's not properly oriented. It would be great if face detect could automatically determine an orientation of the photo that provides a correct face box.
10 votes -
10 votes
-
Beard on face detct doesn't work normaly on people with Beard
There's a problem when we want to know if people have beard or not. I explain:
this picture(https://typeset-beta.imgix.net/elite-daily/2016/06/07173235/fckboy-with-beard.jpg) is a man with beard but when I put it inside Face Detect : here's the result:
"facialHair": {"moustache": 0.1,
"beard": 0.1,
"sideburns": 0.1
}As you can see Face says that this man have no beard.
But when I put this picture(https://previews.123rf.com/images/tverdohlib/tverdohlib1611/tverdohlib161103035/66973877-frown-handsome-bearded-man-with-beard-moustache-and-gray-hair-stylish-hipster-male-in-blue-checkered.jpg) which is another man with beard, the API will send me this:
"facialHair": {"moustache": 0.9,
"beard": 0.9,
"sideburns": 0.6
}So it's working with some pictures and the others aren't working. Maybe it's…
7 votes -
Make confidence level available for face detection
It would be very useful to know how confident the system is that something is or isn't a face when it returns bounding boxes. AWS Rekognition currently does provide this information.
7 votes -
Include Detection process into Identify process
Please consider putting the Detecting process into the Identifying process. This way, we can shorten our code and have less request to your api.
7 votes -
Improve side-face detection
It's quite bad right now
7 voteswe’re always working to improve the detection for different angles, occlusions, etc.
if you have any consistent examples where detection is failing, please share them via the “contact us” link at the bottom of the website https://www.microsoft.com/cognitive-services -
Get a Person Face / Get the picture from the persistedFaceId
It should be great to get the picture that belongs to the persistedFaceId
That allows to manage, especially for removing a picture from the person with a preview of the picture.6 votesThis is not achievable due to legal concern
-
HIPAA compliance
Make the Facial Recognition image data HIPAA compliant to use with sensitive client data.
6 votes -
Provide Flow for PowerApps (Face and Speech prioritized)
Provide prebuilt Face API (and Speech as well if possible) flow capabilities for Microsoft Flow that they can be used in PowerApps on modern devices with microphone and camera capabilities (phone, tablet, desktop computers). They could be used to interact and authenticate with the application.
6 votesPrebuilt flow capabilities for the Face API is now completed:
Further details available here: https://flow.microsoft.com/en-us/connectors/shared_faceapi/face-api/
- Don't see your idea?