Please use this thread to ask for new features
Announcement
Collapse
No announcement yet.
Face Recognition plug-in - Feature Requests
Collapse
This is a sticky topic.
X
X
-
How about support for multiple cameras?
I don't need separate faces profiles happy for that to be shared, but keen to be able to detect across multiple streams. If multiple cameras were possible then I would like the ability to turn on/off the recognition feature for each camera independently (e.g. based on motion detection)
Stretch goal... don't suppose the detection logic could be extended to support vehicle number plate recognition?
-
Originally posted by ShimmerNZ View PostHow about support for multiple cameras?
I don't need separate faces profiles happy for that to be shared, but keen to be able to detect across multiple streams. If multiple cameras were possible then I would like the ability to turn on/off the recognition feature for each camera independently (e.g. based on motion detection)
Stretch goal... don't suppose the detection logic could be extended to support vehicle number plate recognition?
You can also already start/stop the recognition feature by using the Status device Start/Stop controls.
You could use the same training data for all instances by simply copying the images from one directory to another, but I don't think it's a good idea because the recognition algorithm is very sensitive to lights, colors, angle, etc..., so it is always better to record the training data with the same camera and in the same conditions as the ones that the plugin instance is going to use as input.
I'll try to detail all of this in the User Guide
number plate recognition would be possible, but not in my short term todo list.
Comment
-
Originally posted by BIGMACK48 View Postis there anyway to use a clip or video stream taken from the same ip cam from the past and use it to capture face in the manager training data?
Thank you
Greg Mack
Comment
-
Multiple Camera support
I got his working on my cameras!
I used my USB camera to train on my face.
But i have four other IP cameras for security I would like to use.
For me, as one person, I cannot train these cameras because they are all outside at entry doors.
My feature request would be the ability to Train the faces on a jpeg I submit, from that camera.
As an example, my cameras record all motion at these doors. I can take jpeg snap shots of the saved recording, and submit those faces to the trainer.
This would also allow me to train on unknown people who come to my door, I could track when and how often they try my door.
I could also train on my customers. As they walk in, i could have HS send me an email with their name and previous order history or pending orders.
So as I meet them, it is the ultimate customer service where I "remember" them and their purchase history.
I could also train on my employees and get a text when they arrive at work.
Thank you
Thomas
Comment
Comment