Hi and thanks!!! Can you please help me with adding CompreFace to DoubleTake on HomeAssistant, I have 2 issues, AVX not detected and also where to create an api key? I am using Homeassistant on a VirtualBox Linux that is running on a Windows 10 NUC i7 processor that supports AVX
Man everything is getting so good. A.I. has started to move at a fairly good rate after decades. You can install a chatbot on your computer. The open-sourse community is amazing. Science fiction is becoming reality. I for one am excited even if there is a chance of an IRobot/Terminator future. 😂😂😂
Ah amazing I was just about to setup my Frigate + Deepstack integrations based on the last video. Can’t wait to try this, thanks for these amazing videos 🙌
Want to thank you for the great video's you make. Just signed in at your channel like a month ago and Frigate is running smooth together with Double-Take 🥳 Keep up the good work, really appreciate it👍 Greets from the Netherlands
Now I need Frigate/Deep stack for sure. Double Take is complementing the whole solution. But I need a Coral first (like the whole world...). Perfectly described process of the possibilities, I enjoyed watching this and will save it for later! Very good video Lewis!
Thanks from Brazil, your video helped me a lot, I'm using it via Docker through the UNRAID system, in unraid I needed to download docker DOUBLE TAKE and then download the detector separately in docker too (compreFACE), I accessed COMPREFACE through the IP address: port , I registered , I acquired the key inserted in the CONFIG file and everything worked perfect. thanks for the tips, despite being a different system than mine, the automation part was useful to me because I'm still a layman in Home assistant. Thanks.
Hi, as usual, very professional, easy to follow and useful! I've been playing a bit with Doubletake and Frigate, though easy to implement, I am rather disappointed with the speed and accuracy. I use Coral USB TPU for Frigate which makes the object detection run fast but the free available Face detection HA plugins/integrations and detectors are rather older , non maintained versions. I believe current solutions are much accurate and faster . The fact the deepstack is not maintained for few years now and the replacement aiserver is not much better in terms of HA plugin update and speed (does not support Coral TPU at all). Recognition takes almost 100% of resource, takes several to tens of seconds and the result is rather random recognition (I already trained 500 various light condition pictures from the camera in use, but accuracy about 70 % and more often miss and unknown than match for my own pics ...) So to say in use of HA automation this approach is not usable for me. Any better , more modern and reliable solution available ?
This is awesome. I got a bit nervous as I am not a Docker guy but got this all going via the Add-on-store and HACS. Thanks so much. Now for a bit of fine tuning
Great video as always, I really look forward to play with this once I can get my hands on coral. Hopefully double take will mature enough before that. For me its just closing down after mqtt subscription without error even on silly debug level, so no luck there. Actually it wont even start without configuration like it did for you, so maybe they messed up something or did some major change in setup.
and I thought local NVR wouldn't get any better! Super!! I guess the next step now of training faces with high confidence is to drive automations right? would be super cool if home assistant now can open up the door and play "Hi Lewis" on some speaker when you approach haha!
I would like to ask if you could make a video about setting up everything to learn your own models and use them in frigate. I have been trying this. But I can't seem to get it to work. Your tutorials are always so clear. I would love to see a video about learning tflite models and placing them in frigate/home assistant
Very cool video!! Thanks! I have started looking at voice assistants, I'm still haven't added one...Have you looked at "Almond" or other option as a voice assistant vs Alexa or Google?
6:25 you look like a criminal haha, glancing right at the camera great video! I'm not quite ready to get into Frigate, but when I do I'll know where to look. Thanks!
Do you have a server with all your family photos? What software do you use when you have multiple repited photos, to clean and auto organize all this photos (uploaded by 5 members of the family) and to have face recognition and to connect that face recognition like photoprism with this home automation software? Thank you for your videos?
@@EverythingSmartHome Do you have Coral TPU and/or NVIDIA GPU accelerators installed? I guess ideally both as Frigate strongly recommends Coral and Deepstack likes NVIDIA?
Hi, Thanks for another great video. I tried using the combination of Double Take, Frigate, and CompreFace, and I find it much better than DeepStack. However, I have an issue with the CompreFace installation itself. I installed it on my home server, which is based on TrueNAS SCALE. CompreFace is installed as a custom app and runs pretty well until I need to restart it. After a restart, it starts from scratch, losing all data and resetting the API key. It seems I have a problem with setting up persistent storage for the Docker app. I created a dataset as usual and assigned it during the installation, but it appears that the app is not using it. Do you have any idea how to configure the app to use my dataset as persistent storage?
Thanks!! can't find a way to config better the Deepstack. Specially something like this confidence: match: 95 unknown: 30 save: matches: true unknown: true purge: matches: 48 unknown: 12 increase the match level to 95 ... save match and unknown image and specially a way to purge images... or get less images from frigate.! Thanks a lot!!
Could you clarify the host setup you have? You talk about Deepstack and Frigate as Docker containers but Double-take as an add-on. Can these be combined on one machine or are you using more than one? Could Double Take be installed as a container?
Thanks so much for your amazing videos I'm a big fan from Egypt. I need your advice I'm a wheelchair user and want to use face recognition to open an outside gate. my setup is RPI4 on argon one m.2 with 1TB SSD. for HA on docker and Kodi. can I run deepstack or doods, frigate, and double take or should I buy powerful hardware such as Intel nuc?
Hi Lewis, thanks for the great tutorials. Can you suggest something like double-take but for non-face? I'm wanting to recognize my cat so if any cats approach my cat door I can use a servo to block their access. I watched a bunch of your tutorials and managed to get something working but it only works about 60% of the time (intruder cat getting in most nights as its persistent) Any suggestions would be much appreciated.
Hi Lewis, is it possible to use an esp32 cam for this? It is inexpensive, local and small. But I don’t know if the quality is good enough. I don’t want to use my normal cameras for this, I wan’t them to be disabled when I am home. Or maybe you know of another inexpensive and small camera that can be used locally and is good for this. Thanks!
Excelent video, thanks! I´m using it for a month but i only can see the matches from the last 5 days, do you know if there is a way to extend this to a month or more?
**spoiler** stupid question ahead **/spoiler** When our cameras are actually ESP32-CAMs attached through ESPHOME, should we still use Frigate, if so how? or would they be already integrated so maybe we even do not need to use Frigate at all? But I understand Frigate triggers the motion / person detection at first place. A bit confused, any help would be highly appreciated 🤩
Thanks for this brilliant video... But I only came to post as I love the t-shirt 🤣... 🤘 (This video was also excellent, I shouldn't detract from that 🤗)
Hi Lewis You mentioned in the video, that it would be possible to avoid getting double pictures (From both MQTT and Snapshot) in the DoubleTake Gui. I have been searching around to find out how to keep only the snapshot image and discard the mqtt image for double take. Is this configuration something to be done in Frigate or DoubleTake? Cheers Anders
hello, thank you for the video. I have a error : error: TypeError: deepstack process error: Cannot read properties of undefined (reading 'flatMap'). do you know ?
hi. can i use a different port? my 3000 uses euph. after changing to 3001 in docker compose double take it doesn't show the dockerhost: 3001 configuration page. there are no errors
Love your channel, superb videos and learning loads! I've just setup an Amcrest AD410 doorbell with Frigate, Deepstack and Doubletake because of this video. It's super super cool, I've put an order in for a Coral accelerator now too! Quick question, is there any way to feed the trained names from Doubletake into Frigate? At the moment under "Labels" it just gives me "Person" but it would be cool to have the ability to filter events based on who it detected! Thanks for the great content, I've subscribed! Conor
@@LuisPeris Hi Luis, the Coral detector only benefits the object detection on Frigate - I don't believe it adds any benefit to doubletake at the moment.
Do you think all of this can run on an rpi4? Currently running frigate on my rpi4 on HASSIO without a coral but looking in to getting one when possible and its pretty meh.
Hi and thanks!!! Can you please help me with adding CompreFace to DoubleTake on HomeAssistant, I have 2 issues, AVX not detected and also where to create an api key?
I don't see how you can use the state of. sensor.double_take_xxx to trigger, since the state is the camera name. If the same person walked in front of the same camera. the state wouldn't change. Some details of the sensor would change like the timestamp, filename etc but not the state. Actually that seems like a very strange choice for the state. If the state reverted to OFF after a certain period of time you would have a state change of multiple appearances but the state appears to stay as the last occurence indefinitely.
@@EverythingSmartHome My point has nothing to do with the examples. The question is how to handle the fact that there is no state change with multiple recognitions. I usually use node red instead of HA to actually "DO" something (old javascript coder). I am thinking that after the first recognition, set the state of the entity to "processed" manually. That way when the next occurence happens there will be a state change to trigger things. The other alternative is to trigger off the mqtt message, wait a couple of seconds and see if there is a match found by deep stack. It just seems to me that using a camera name as the state doesn't make much sense.
Thank you very much for a crystal clear explanation. Nowadays I'm struggling with Frigate Add-on as since I updated HA to 2021.12 detection sensor stopped working. Any lead to solve it? Thank you in advance
so lets say the camera could recognize you, can it do some scenarios when it recognize your face?, like : when the lewis face regognite set the temperature to 25 delicious degrees or some other scenarios?
Great work Lewis! I have Frigate and Double Take up and running. I have added a face recognition folder for myself, trained it, and the sensor shows up for unknown and myself. I have since added other folders for more people and double take is not adding more sensors. How do I get the sensor.double_take_XXXX for each person I train?
@@EverythingSmartHome I guess that makes sense. I have trained on some detected faces and still not seeing the sensor. It finally showed up when deepstack recognized the face correctly. I guess I will have to line up my family in front of a camera like a line up to get the sensors made. :) Appreciate the response and love your videos.
@@EverythingSmartHome I had the impression that Double Take is similar to Deepstack UI for training detection, but didn't have a chance to go deeper to have a better conclusion. Will definitely try both.
5/24/23: I tried to install the add-on, but when I pasted the URL into the place to add a repository, I got a message that it's "not a valid add-on." Even though I proceeded as if it were (valid, that is), I never saw it in the list, so I wasn't able to install the Double-Take Hassio addons. Got any ideas? Thanks in advance! (I sure hope I get a response!)
I am getting matches after trains but the only sensor (other than the cameras) is double_take_unknown. Shouldn't I have double_take_don or do I have to add something to config to pick that up?
You will see if you look in the attributes of the sensor it will tell you why it wasn't matched - usually it's because the box wasn't large enough, this is an option you can configure in the double take settings to tweak it to your camera settings
@@EverythingSmartHome Found it. Thanks. Any reccomendations for a POE cam with either a very high zoom or narrow FOX. Idea would be that I slow down as I come up my driveway and if it recognizes me triggers the garage door to open. Mount point is probably 30 ft from vehicle and most cams are so wide that a "head shot" at 30 ft. is very few pixels.
It seems (and I'm new to this) that the DoubleTake repository also has Deepstack addons listed. Why not use one of these Deepstacks inside HA rather than an external one?
@@EverythingSmartHome LOL! Gotcha. It seems to connect and show a green dot. (Have to use port 5001.) I'm gonna try with the internal Deepstack and see how it goes. Thanks for the quick reply. Fantastic videos! Keep up the great work; you're an amazing resource and your hard work is much appreciated.
What happens if you walk in front of the camera with a still picture of yourself? Does it recognizes that you're trying to fool the system? I'm wondering the security side...
Yes, I've said several times before never to use this as your only mechanism of security for alarms etc, but rather to use it as an additional measure and combine it with other factors. Or just use it for enhancing your CCTV systems
Installing the add-on killed HA for me. Looks like it pulled and installed deepstack container which I had already installed on a different vm. Removing that and rebooting seems to have gotten me back up and running. I am thinking maybe you need to install the config with urls before you start.
User error. I must have clicked on install deep stack instead of double take. Note to self: have that second cup of coffee so you are awake when installing
I've been triggered by my frigate , for I haven't had chance to install any of my cameras due to the non-stops raining / snowing the last few days , I pointed my camera in the living room to look for cat / dog / bicycle / birds / cars / horse , which unless a car crashes into the living room it shouldn't find any and frigate object detection thinks I'm 84% a dog which is higher most times than being a person . So time to change my name to rover !!!!!
If you get a coral then you may get away with it, you may want to check the requirements for something like Compreface to see if that can work on the Pi!
Face Recognition is getting so good! 😱
Hi and thanks!!! Can you please help me with adding CompreFace to DoubleTake on HomeAssistant, I have 2 issues, AVX not detected and also where to create an api key? I am using Homeassistant on a VirtualBox Linux that is running on a Windows 10 NUC i7 processor that supports AVX
Man everything is getting so good. A.I. has started to move at a fairly good rate after decades. You can install a chatbot on your computer. The open-sourse community is amazing. Science fiction is becoming reality. I for one am excited even if there is a chance of an IRobot/Terminator future. 😂😂😂
@@effectentertainment7882 And now you can watch your security cameras in 4K in Apple Vision Pro!
Great video! Frigate, doubletake, and deepstack have made automations with face recognition so much easier!
Nice 🙌
Ah amazing I was just about to setup my Frigate + Deepstack integrations based on the last video. Can’t wait to try this, thanks for these amazing videos 🙌
Nice one, have fun!
Want to thank you for the great video's you make. Just signed in at your channel like a month ago and Frigate is running smooth together with Double-Take 🥳
Keep up the good work, really appreciate it👍
Greets from the Netherlands
Amazing, thank you so much for the kind words!
Now I need Frigate/Deep stack for sure. Double Take is complementing the whole solution. But I need a Coral first (like the whole world...). Perfectly described process of the possibilities, I enjoyed watching this and will save it for later! Very good video Lewis!
Thanks! Good luck with a coral, hope you find one soon!
Just got Frigate working and detecting people, cars and dogs. Time to take it to the next level. Thanks for the awesome video.
Love the shirt. I love seeing MCM merch out in the wild.
Love me some MCM!
IN THE BIN!
Thanks from Brazil, your video helped me a lot, I'm using it via Docker through the UNRAID system, in unraid I needed to download docker DOUBLE TAKE and then download the detector separately in docker too (compreFACE), I accessed COMPREFACE through the IP address: port , I registered ,
I acquired the key inserted in the CONFIG file and everything worked perfect. thanks for the tips, despite being a different system than mine, the automation part was useful to me because I'm still a layman in Home assistant. Thanks.
Hi, as usual, very professional, easy to follow and useful! I've been playing a bit with Doubletake and Frigate, though easy to implement, I am rather disappointed with the speed and accuracy. I use Coral USB TPU for Frigate which makes the object detection run fast but the free available Face detection HA plugins/integrations and detectors are rather older , non maintained versions. I believe current solutions are much accurate and faster . The fact the deepstack is not maintained for few years now and the replacement aiserver is not much better in terms of HA plugin update and speed (does not support Coral TPU at all). Recognition takes almost 100% of resource, takes several to tens of seconds and the result is rather random recognition (I already trained 500 various light condition pictures from the camera in use, but accuracy about 70 % and more often miss and unknown than match for my own pics ...) So to say in use of HA automation this approach is not usable for me. Any better , more modern and reliable solution available ?
This is awesome. I got a bit nervous as I am not a Docker guy but got this all going via the Add-on-store and HACS. Thanks so much. Now for a bit of fine tuning
Nice job 🙏
I'm so into Frigate right now, so this was perfectly timed!
Awesome, enjoy!
Great video Lewis. DT+Frigate are an incredible combination. Amazing automations can be done...:)
Thank you sir!
Thanks. Been hanging for this. Goodbye early nights for the foreseeable future!
Enjoy! 😅
Awsome! Looking forward to get finaly my cam and configure this 👌🏽 great job of the developer. A long time test would also be great to see
Nice one, enjoy!
Can you pls make a video on CompreFace and doubletake?
Hahah just finished movin my hikvision cams from my expensive hikvision NVR to frigate ;) this is just awesome !!!
Thanks for the vid !
Haha nice, enjoy!
Great video as always, I really look forward to play with this once I can get my hands on coral.
Hopefully double take will mature enough before that. For me its just closing down after mqtt subscription without error even on silly debug level, so no luck there.
Actually it wont even start without configuration like it did for you, so maybe they messed up something or did some major change in setup.
I was waiting for your guide! Thanks!
Glad to provide it!
and I thought local NVR wouldn't get any better! Super!!
I guess the next step now of training faces with high confidence is to drive automations right? would be super cool if home assistant now can open up the door and play "Hi Lewis" on some speaker when you approach haha!
Haha that would be amazing...and creepy to some people 😂 do it! And yes getting confidence higher would be a must
can you make an update video on NVR and face recognition on home assitant in 2023, please.
In the bin! 😁
Too funny 😂
Thanks for sharing mate, awesome as usual
Thanks man!
Hi Lewis, brilliant explanation. Thank you!
Thanks, glad you liked it!
Since DeepStack is archived, how do we install this now?
your videos are very good and have a wide message thank you
Thanks Emilia!
really nice, thanks for sharing!!!! awesome as always
Thanks for watching as always! 😅
I would like to ask if you could make a video about setting up everything to learn your own models and use them in frigate. I have been trying this. But I can't seem to get it to work. Your tutorials are always so clear. I would love to see a video about learning tflite models and placing them in frigate/home assistant
Very cool video!! Thanks! I have started looking at voice assistants, I'm still haven't added one...Have you looked at "Almond" or other option as a voice assistant vs Alexa or Google?
In the BIN!!!
6:25 you look like a criminal haha, glancing right at the camera
great video! I'm not quite ready to get into Frigate, but when I do I'll know where to look. Thanks!
That's me, just a common criminal 😂
Unexpected MCM ... IN THE BIN!
Nice to see a mighty car mods t-shirt :)
Love me some MCM!
Do you have a server with all your family photos? What software do you use when you have multiple repited photos, to clean and auto organize all this photos (uploaded by 5 members of the family) and to have face recognition and to connect that face recognition like photoprism with this home automation software? Thank you for your videos?
Great video, Lewis! Thanks for sharing this with us. Quick question. On what are you running everything?
Thanks! I use a PowerEdge server at the moment 😅
@@EverythingSmartHome Do you have Coral TPU and/or NVIDIA GPU accelerators installed? I guess ideally both as Frigate strongly recommends Coral and Deepstack likes NVIDIA?
Hi,
Thanks for another great video. I tried using the combination of Double Take, Frigate, and CompreFace, and I find it much better than DeepStack. However, I have an issue with the CompreFace installation itself.
I installed it on my home server, which is based on TrueNAS SCALE. CompreFace is installed as a custom app and runs pretty well until I need to restart it. After a restart, it starts from scratch, losing all data and resetting the API key. It seems I have a problem with setting up persistent storage for the Docker app.
I created a dataset as usual and assigned it during the installation, but it appears that the app is not using it. Do you have any idea how to configure the app to use my dataset as persistent storage?
Do you still need the DeepStack-Face config when using DoubleTake or does it comes with its own config?
And what about object detection?
On what kind of hardware is this run ? Do I need a Jetson Nano for this ?
Jako is my buddy!! woo Jako!
Hahaha!
Great content. Thank you!
Cheers Jack!
Thanks!! can't find a way to config better the Deepstack. Specially something like this confidence:
match: 95
unknown: 30
save:
matches: true
unknown: true
purge:
matches: 48
unknown: 12
increase the match level to 95 ... save match and unknown image and specially a way to purge images... or get less images from frigate.! Thanks a lot!!
Could you clarify the host setup you have? You talk about Deepstack and Frigate as Docker containers but Double-take as an add-on. Can these be combined on one machine or are you using more than one? Could Double Take be installed as a container?
Thanks so much for your amazing videos I'm a big fan from Egypt. I need your advice I'm a wheelchair user and want to use face recognition to open an outside gate. my setup is RPI4 on argon one m.2 with 1TB SSD. for HA on docker and Kodi. can I run deepstack or doods, frigate, and double take or should I buy powerful hardware such as Intel nuc?
Where is the sample frigate yaml u said is available in the description?
Hi Lewis, thanks for the great tutorials. Can you suggest something like double-take but for non-face?
I'm wanting to recognize my cat so if any cats approach my cat door I can use a servo to block their access. I watched a bunch of your tutorials and managed to get something working but it only works about 60% of the time (intruder cat getting in most nights as its persistent)
Any suggestions would be much appreciated.
Can't you use a tag in the collar to detect if your cat is within range?
Some cat doors can detect an id chip, which nowadays usually replace the identification tatoo...
I see that the double-take repo has a deepstack add-on too. Would that install make life a bit easier?
Yeah certainly could do, I didn't include or use it in the video since it was released after I started planning and shooting this video
I tried with the addon but could not get Doubletake to show any pictures, no idea what I'm doing wrong
What would be a good setup im 2024?
I see that double take was not updated in almost 2 years, so it's unlikely to still work correctly.
Just need to add the Edge TPU and make it radically faster! Hope this is in work somewhere.
Hi Lewis, is it possible to use an esp32 cam for this? It is inexpensive, local and small. But I don’t know if the quality is good enough. I don’t want to use my normal cameras for this, I wan’t them to be disabled when I am home. Or maybe you know of another inexpensive and small camera that can be used locally and is good for this. Thanks!
Got it working within a few minutes. How do I train a model with 2 people in the frame? It draws two boxes but reverses the detected face
good afternoon. the error 502: Bad Gateway comes out in double take. I did everything according to your video. fix what?
Can i train the AI for other things? Such as fire detection at home?
Excelent video, thanks! I´m using it for a month but i only can see the matches from the last 5 days, do you know if there is a way to extend this to a month or more?
**spoiler** stupid question ahead **/spoiler**
When our cameras are actually ESP32-CAMs attached through ESPHOME, should we still use Frigate, if so how? or would they be already integrated so maybe we even do not need to use Frigate at all? But I understand Frigate triggers the motion / person detection at first place. A bit confused, any help would be highly appreciated 🤩
Thanks for this brilliant video... But I only came to post as I love the t-shirt 🤣... 🤘
(This video was also excellent, I shouldn't detract from that 🤗)
Hahahaha thanks! As long as I get the comment to help the almighty algorithm, I'm fine with that 😂 plus MCM are 🐐
Hi Lewis
You mentioned in the video, that it would be possible to avoid getting double pictures (From both MQTT and Snapshot) in the DoubleTake Gui. I have been searching around to find out how to keep only the snapshot image and discard the mqtt image for double take.
Is this configuration something to be done in Frigate or DoubleTake?
Cheers
Anders
Hey, did you ever find an answer for this? just struggling with it myself...
Hi, do I need the CORAL USB to go together with this?
Hello,
First thank for the video is great,
Question do i need to install deepstack?
hello, thank you for the video. I have a error : error: TypeError: deepstack process error: Cannot read properties of undefined (reading 'flatMap'). do you know ?
I cant start Double take because port 3000 is occuppied, how to change port?
Just curious. What device do you run your Home Assistant on?
hi. can i use a different port? my 3000 uses euph. after changing to 3001 in docker compose double take it doesn't show the dockerhost: 3001 configuration page. there are no errors
Deepstack it doesn work in the Double take configuration, it appears econnrefused in the log of double take.
i have frigate doing person and dog detection.
doubletake now recognizes my dog at the door.
amaizing work cheers
Thanks buddy!
Love your channel, superb videos and learning loads! I've just setup an Amcrest AD410 doorbell with Frigate, Deepstack and Doubletake because of this video. It's super super cool, I've put an order in for a Coral accelerator now too!
Quick question, is there any way to feed the trained names from Doubletake into Frigate? At the moment under "Labels" it just gives me "Person" but it would be cool to have the ability to filter events based on who it detected!
Thanks for the great content, I've subscribed!
Conor
Thanks for the support! I don't think it's possible to do that at the moment unfortunately
@@EverythingSmartHome support for this is finally here under Frigate 0.11.0 🥳
Hi Conor, do you have the double take file config for coral detector? thanks in advance
@@LuisPeris Hi Luis, the Coral detector only benefits the object detection on Frigate - I don't believe it adds any benefit to doubletake at the moment.
@@ConorJay then how do i configure double take? Do i need another hardware similar as google coral?
Cannot add the hacs repository any longer?
I’m going to sound dumb (I’m a little new to security setups) but does this work with Ubiquiti equipment?
Yes, it does.
Can this work with other objects, like car types?
Do you think all of this can run on an rpi4? Currently running frigate on my rpi4 on HASSIO without a coral but looking in to getting one when possible and its pretty meh.
Great content thanks
Thanks for watching!
Hi and thanks!!! Can you please help me with adding CompreFace to DoubleTake on HomeAssistant, I have 2 issues, AVX not detected and also where to create an api key?
any idea why the double take add on won't start for me? config.yaml is empty and I get no logs
I don't see how you can use the state of. sensor.double_take_xxx to trigger, since the state is the camera name. If the same person walked in front of the same camera. the state wouldn't change. Some details of the sensor would change like the timestamp, filename etc but not the state. Actually that seems like a very strange choice for the state. If the state reverted to OFF after a certain period of time you would have a state change of multiple appearances but the state appears to stay as the last occurence indefinitely.
Sure, it was just a really quick example, not really to copy word for word but just a basic overview of some sensors etc
@@EverythingSmartHome My point has nothing to do with the examples. The question is how to handle the fact that there is no state change with multiple recognitions. I usually use node red instead of HA to actually "DO" something (old javascript coder). I am thinking that after the first recognition, set the state of the entity to "processed" manually. That way when the next occurence happens there will be a state change to trigger things. The other alternative is to trigger off the mqtt message, wait a couple of seconds and see if there is a match found by deep stack. It just seems to me that using a camera name as the state doesn't make much sense.
You could simply use a template and look at the attributes instead since those change, or yes the Mqtt message is also a great way to do it
Thank you very much for a crystal clear explanation. Nowadays I'm struggling with Frigate Add-on as since I updated HA to 2021.12 detection sensor stopped working. Any lead to solve it? Thank you in advance
Thanks for watching! I haven't had any issues yet with mine but I'll need to look into it
get this error Cannot read properties of undefined (reading 'id')
is there anyway i can use my intel neural compute stick 2 with Dee-stack i have it plugged in just hoping its working, thank you for the great videos
so lets say the camera could recognize you, can it do some scenarios when it recognize your face?, like : when the lewis face regognite set the temperature to 25 delicious degrees or some other scenarios?
Yes of course. With home assistant you can trigger anything based on who's recognised.
Any doorbell camera you can use this with?
Great work Lewis! I have Frigate and Double Take up and running. I have added a face recognition folder for myself, trained it, and the sensor shows up for unknown and myself. I have since added other folders for more people and double take is not adding more sensors. How do I get the sensor.double_take_XXXX for each person I train?
Hey Greg thanks! The sensor won't show up until it actually sees a match in the camera, have you done that yet?
@@EverythingSmartHome I guess that makes sense. I have trained on some detected faces and still not seeing the sensor. It finally showed up when deepstack recognized the face correctly. I guess I will have to line up my family in front of a camera like a line up to get the sensors made. :) Appreciate the response and love your videos.
Good video
Can you read our minds!!!!!!
Maaaaybe 🔮😂
Double posting sorry, can you comment on the requirements for a system hosting only HA + Frigate, Deepstack and Doubletake
Hey thank you for sharing! This is gold. Question: can Deepstack UI also be used for this same purpose?
Hello! I'm not sure what you mean sorry?
@@EverythingSmartHome I had the impression that Double Take is similar to Deepstack UI for training detection, but didn't have a chance to go deeper to have a better conclusion. Will definitely try both.
@@rsfurlan90 MostlyChris also made a quite detailed video ua-cam.com/video/5OPOAusvo8I/v-deo.html
Amazing.
Not so amazing: my dash cam takes your English as commands... It's been snapping pictures non-stop for 9 minutes.
Thanks Lewis!
Wait what?! 😂 How?!
@@EverythingSmartHome I don't know, it just kept acknowledging the commands and snapping pictures. Thanks for the videos! 📸
Xiaomi, do I say more? 🤣
Hahaha must be my pesky Scottish accent, can't understand me 😂
what cameras would i need for this? Can i use my eufy cameras ?
5/24/23: I tried to install the add-on, but when I pasted the URL into the place to add a repository, I got a message that it's "not a valid add-on." Even though I proceeded as if it were (valid, that is), I never saw it in the list, so I wasn't able to install the Double-Take Hassio addons. Got any ideas?
Thanks in advance! (I sure hope I get a response!)
Probably double take is outdated, it wasn't updated since 2022.
I am getting matches after trains but the only sensor (other than the cameras) is double_take_unknown. Shouldn't I have double_take_don or do I have to add something to config to pick that up?
You will see if you look in the attributes of the sensor it will tell you why it wasn't matched - usually it's because the box wasn't large enough, this is an option you can configure in the double take settings to tweak it to your camera settings
@@EverythingSmartHome Found it. Thanks. Any reccomendations for a POE cam with either a very high zoom or narrow FOX. Idea would be that I slow down as I come up my driveway and if it recognizes me triggers the garage door to open. Mount point is probably 30 ft from vehicle and most cams are so wide that a "head shot" at 30 ft. is very few pixels.
It seems (and I'm new to this) that the DoubleTake repository also has Deepstack addons listed. Why not use one of these Deepstacks inside HA rather than an external one?
It was only literally created after I started planning this video and only been out for 2 days by the time this video was released
@@EverythingSmartHome LOL! Gotcha. It seems to connect and show a green dot. (Have to use port 5001.) I'm gonna try with the internal Deepstack and see how it goes. Thanks for the quick reply. Fantastic videos! Keep up the great work; you're an amazing resource and your hard work is much appreciated.
No worries glad they are useful!
What happens if you walk in front of the camera with a still picture of yourself? Does it recognizes that you're trying to fool the system? I'm wondering the security side...
Yes, I've said several times before never to use this as your only mechanism of security for alarms etc, but rather to use it as an additional measure and combine it with other factors. Or just use it for enhancing your CCTV systems
How do I select the storage to my local HDD?
Another great one. Does this work with blue iris??
I don't believe so at the moment!
Can you still get the sensors in home assistant for the automations if you run Double Take in a container outside of home assistant?
Yes because it uses MQTT auto discovery, just make sure you have that enabled in the Home Assistant MQTT integration
@@EverythingSmartHome Awesome! Thank you for another great video!
Installing the add-on killed HA for me. Looks like it pulled and installed deepstack container which I had already installed on a different vm. Removing that and rebooting seems to have gotten me back up and running. I am thinking maybe you need to install the config with urls before you start.
User error. I must have clicked on install deep stack instead of double take. Note to self: have that second cup of coffee so you are awake when installing
Hahaha glad you figured it out 😂
I've been triggered by my frigate , for I haven't had chance to install any of my cameras due to the non-stops raining / snowing the last few days , I pointed my camera in the living room to look for cat / dog / bicycle / birds / cars / horse , which unless a car crashes into the living room it shouldn't find any and frigate object detection thinks I'm 84% a dog which is higher most times than being a person .
So time to change my name to rover !!!!!
Is a raspberry powerful enough to run this?
this is cool
This is so awesome. Good-bye nest camera.
Haha enjoy Alex!
@@EverythingSmartHome this is the future. 100% local smart home! That’s my goal
Why do I hear a very faint alternating low-high tone in the background? It's driving me crazy! lol
How much power do I need for person and face detection? At the moment I am using an RPi 4 + SSD 😊
If you get a coral then you may get away with it, you may want to check the requirements for something like Compreface to see if that can work on the Pi!
@@EverythingSmartHome what is coral?😊 What hardware do you use? 🤔
@@Chris-cj5hz look at his previous videos :)
All right! 😁
In a Hurry?