Robot Not a Dev? Pre-order Now
Not a Dev?

Misty Community Forum

Face detection observation/Issue maybe

Just a general question about face recognition / detection:

Face recognition training works great, Luna (My daughter changed Misty’s the name) knows who I am once I am in view but when I switch to facial detection, Luna no longer recognizes me but does recognize that it has detected a face. It seems like the skill is working as defined, Luna detects a face, but in the console the personName variable comes back as “unknown person”. Is this the correct return value? If it is and Luna cannot recognize learned faces during the face detection skill, why have a variable that implies it can return a learned person, personName? Will Luna be able to detect a face and the person if it learned in the future?

Thanks and everything else is great so far aside from the jerkiness turning on hardwood floors.

Anthony

Hi @stackchain - curious as to if you are using a skill for this or trying via the Command Center interface.

If you have taught Luna your face (and named yourself when doing that), her default behavior should be to return that name in place of personName. This is an existing functionally that should be available.

In order to start Misty recognizing the faces she sees, you need to use the misty.StartFaceRecognition() method.

If you use the misty.StartFaceDetection() method instead, then the system still returns messages from FaceRecognition events, but my understanding is those messages won’t return the name of the person, even if it’s someone Misty knows.

Hi @Ben ,

I was at the demo in New York, I had a good time. Thanks for coming!

I was doing everything in the command center. I first taught her using face training and that worked fine when I tested it with FaceRecognition. The console output:

{eventName: "FaceRecognition", message: {…}}
eventName: "FaceRecognition"
message:
bearing: -6
created: "2019-09-12T21:04:47.8278743Z"
distance: 55
elevation: 0
personName: "anthony"
sensorId: "cv"
trackId: 822
__proto__: Object
__proto__: Object

When I switched to FaceDetection this is the output:

{eventName: "FaceDetection", message: {…}}
eventName: "FaceDetection"
message:
bearing: 5
created: "2019-09-12T21:06:02.3154411Z"
distance: 66
elevation: -3
personName: "unknown person"
sensorId: "cv"
trackId: 891
__proto__: Object
__proto__: Object

I was just wondering if this is normal. If i write skills that involve getting identities, I will have to use face recognition with face detection??.. I haven’t gotten that far yet, It’s only been a day so far…

Thanks Ben,
Anthony

@Ben,

You actually answered my question in your first response. I read through it to fast the first time…

Thanks Ben,
Anthony

Awesome that Misty recognizes you now !!

Yes, they are supposed to work that way. When you are writing a skill you could choose to use either one based on what you need.

Key advantage of using FaceDetection as opposed to Face Recognition is that — Speed of Face Detection is faster than FaceRecognition (a small difference in latency - yet matters).

Here are some Skill Examples to get started on Face Detection and Face Recognition

1 Like

This is completely understandable and i can see how this would be adventitious for some skills…
Thanks @cp_sridhar

Anthony

1 Like