![]() For DeepStack we need to enter our DeepStack URL, and if you are using an API key then you can specify that here too, I’m not so I will leave that blank: In this case we are using DeepStack for this, but if you want to use another detector either Compreface or Facebox then you will want to slightly tweak this config here. Url: The last thing to add is our detector, the detector is what will process our facial recognition from Frigate, once a person is found. Next enter the URL of your Frigate server and if you are running Frigate on Home Assistant then this will just be the same IP or hostname as your Home Assistant server, with port 5000 at the end. This needs to be the the same MQTT broker that your Frigate server is connected too as this is how it communicates with Frigate to start the facial recognition: mqtt: Open up the Double Take UI and you will land at the main dashboard – you will see that it says no files found which is expected as we haven’t yet told it where Frigate is located yet – go up to the top, head over to the config tab and then we need to configure our MQTT with our Frigate and DeepStack details which is pretty straight forward.įirst enter the MQTT broker details and making sure to include the IP address, username and password for your MQTT details. Go ahead and hit the start button, since all of the config is done inside the Double Take UI. Once installed, you can tick watchdog, auto update and show in sidebar options if you wish. Then find the newly added Double Take entry and hit the install button. ![]() Paste in the this link which will add a new repository for the Double Take add-ons. Head over to supervisor then add-on store and hit the three dots and click repositories. Also, if you want to follow along with this guide, you are going to need to have Frigate already installed and working.Īlright, so now that we have Frigate and DeepStack up and running, lets get into the install! Now, as a little side note, literally just a week or two ago, Double Take officially added the install option for running as a Home Assistant add-on – making our lives much easier! If you don’t use Home Assistant add-ons then it is straightforward if you want to run it as a docker container. We are going to be using DeepStack for this guide since we’ve already covered the install process (again, no article sorry…OK I was REALLY behind) for that in the past but if you want to use any of the other detectors then the configuration should be pretty similar. ![]() It is also worth mentioning that Double Take works with Frigate as the NVR which handles the recording of footage, snapshots and its handling the object detection – looking for a person in the camera image when a person is detected.ĭouble Take will then pass that image into what is called a detector for facial recognition and, as mentioned earlier, Double Take supports 3 different detectors: DeepStack, Compreface and Facebox. Well… Double Take is going to take care of that whole process for us with a nice clean UI to manage it all, right from inside Home Assistant. So, with the Frigate and DeepStack setup, we used to get Frigate to detect a person and then when that happened, we would use a Home Assistant automation to get DeepStack to run facial detection on that image. The other thing that wasn’t the easiest to do was run facial detection automatically after a person was detected. This makes it much easier for anyone to use than ever before and that was one of the challenges with DeepStack.ĭeepStack wasn’t the easiest nor most user friendly thing to train faces for but Double Take helps us to easily solve that issue. Double Take supports several detectors, integrates with Home Assistant as well as MQTT and, the cool bit is that you can actually train and un-train faces right from the UI. Whilst both of those systems are great – to get it working required little more manual intervention and they felt disconnected from each other, unfortunately.Įnter Double Take: a unified interface to combine the powers of both object detection and face recognition in one easy to access place, with the ability to train facial recognition images in seconds right from inside of Home Assistant! Video:įirstly, lets take a look at what Double Take is along with some of the features and benefits that it gives you: Double Take is an open source project created by David Jakowenko designed to be a unified interface for processing and training images for facial recognition. A few months ago I covered the ultimate NVR install (sorry no article on that, I got a bit…behind on posting!) which covered combining the object detection and recording capabilities of Frigate, with the face recognition of DeepStack to give you the best of both worlds.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |