@ravensteijn, thank you for the compliments. This is a bug I missed. I’ll push up a fix today and let you know when it’s ready.
@ravensteijn I pushed an update and new release that should fix this. docker pull jakowenko/double-take:0.9.1 should include this fix. Let me know if it resolves your issue.
Hi guys,
interesting product.
Is there any manual for noobs how to install it.
I use NUC with docker.
Thanks a lot
Hey @thisIO. The Docker section in the README should get you started. You can also join our Discord and I can help you more on there.
it’s working! thnxs for the fast fix
1 Like
Thanks a lot, i have insttalled it and trained some persons. I have deepstack in my setup.
Bu i have nothing on matches page
I can get deepstack working only if i trigger image_processing.scan from node-red.
So how do i get it scanning images from double take and show them in UI?
Thanks a lot!
Jump on our discord and I can help you more. What source are you using for your images? If you’re using Frigate, then you can drop that image_processing.scan completely.
Hey Jako, any intent on turning this into an add-on?
Would be pretty cool to see an addon.
@omayhemo, @danbutter it’s something I’ve considered, I need to do a little research on what’s involved. Is there anything outside of the current feature set you were hoping from an add on? Or is just the ease of installation the main idea?
@Jako addon it’s just a docker that only works in hassos. No additional features only convenience of being in the hassos supervisor and one click installation.
Nevertheless hassos users can still install your docker as a normal docker as well.
For all these “addons” like frigate, doubletake, compreface, etc. I prefer to not run them inside hassos due to memory, cpu and overall resource usage from within hassos. My hassos is busy enough :-).
Also from a “all-in-one-box” principle I prefer to not have it that way (for above mentioned “AI” stuff).
I would like to avoid managing another container outside of HA. It eases configuration and installation as well as updates.
Same here except I don’t run anything in docker on my own. Only haos addons which I don’t have to fully manage.
The machine running my haos has plenty of power so I don’t worry about that.
If it is too involved then no biggie…don’t do it just for me. It is just a would be nice to see type of thing.
@Jako This is a great integration however deepstack detection isn’t very consistent and I have a few questions.
- I’ve trained the faces of my family members but the Matches page shows different results. It sometimes shows a person identified as the wrong person.
What’s the best way to handle this?
Do I select the item on the Matches page and delete or should I highlight the person and then select the correct person in the top left drop down and select Train?
- Feature request (unless I’m missing how to do it)
Allow the choosing of a person in the drop down box and then displaying all images used for that person.
For example:
on the Train page: choose person in Dropdown box… show another option +Upload / Sync / Untrain / Show <------- show would show all images for the selected person
or
on the Matches page: choose person in Dropdown box… show another option Train / Show <------- show would show all images for the selected person
@juan11perez @sender @omayhemo @danbutter thanks for the feedback on the add-on. It does seem pretty straightforward and for some people it makes sense since it makes installation/updates a lot easier.
I’ll start working on support for that next. I don’t run a supervised version HA currently, so need to get a new install running to develop this.
I’ve currently been adding in support for login to make using this even more secure and allows people to use a reverse proxy and not have to worry about adding in their own auth.
All the API endpoints will then require authentication, so I’m also adding in support to create long lived access tokens similar t what HA does. More to come on this when it’s ready.
4 Likes
Hey @surge919,
False positives are possible and it varies depending on the detector used and the images used in training. I’ve been trying to only use higher quality photos for training and uploading selfies or pictures from my phone on the train page.
Deleting images from the matches page just removes that record from the page. It does not effect anything for the trained user.
If you have good quality photos on the match page for a user or incorrect user you can select those and train it towards the user you want. Try to use images where the face is clear and large. I originally trained using a lot of images where the faces from my cameras were blurry or hard to see and that lead to more false positives.
I’ve also increased my confidence.match value to 90 to try to decrease the amount of false positives I get.
I like your idea for the feature request. I was thinking of something similar as well so users can easily see the images for a specific person. I’m finishing up a login module for this project and then was going to look into turning it into an HA addon. I’ll try to work on a solution for your idea as well!
hi, how do you make the config file in the same directory? What directory? Double-take conatiner
directory? can you explain a little further please. Docker noob here…
Hey @Eddie1974, if you mount the volume from the container to your host it will put the config.yml file in there for you if it doesn’t exist.
In the below example it would be located at ${PWD}/.storage/config.yml. You can then use the UI to edit it.
version: '3.7'
services:
double-take:
container_name: double-take
image: jakowenko/double-take
restart: unless-stopped
volumes:
- ${PWD}/.storage:/.storage
ports:
- 3000:3000