Facial recognition & room presence using Double Take & Frigate

I noticed the same. Wonder how the system works!

This is an example if you integrate codeproject/ai-server into a existing docker-compose file. This is not directly intended to work with home-assistant OS / supervised. Please also note that <data> needs to be specified too. Please adapt it to you specific configuration.

I have tested it on Intel (amd64) basis, which works fine so far.

I know, both @Bigmack410 and @imranulbari need help with this. I don’t think they understand what docker is and where to add those lines of code. But I could be wrong, of course.

You are correct. Could explain to me in more detail how you got this to work. I just run a version of codeproject on windows11 machine . I don’t used docker. I now have blue iris on different windows11 machine talking to codeproject for face recognition on another window pc. Can the same be done with double take.

Thanks

Greg

please find frigate configuration from home assistant add-ons and do the following. It does work for me.

image

1 Like

Could you explained in more detail how you set this up. I little confused at this time. Any help will be helpful.

I got frigate working the other day by changing configuration file. But I still getting errors in double take like what was posted on web site before.

The error I getting now:

23-04-24 08:25:26 info: processing frontyard: 1682339122.491281-0z1h7v
23-04-24 08:25:26 error: TypeError: aiserver process error: Cannot read properties of undefined (reading ‘recognize’)
at module.exports.recognize (/double-take/api/src/util/detectors/actions/index.js:4:24)
at Object.module.exports.process (/double-take/api/src/util/process.util.js:160:28)
at Object.module.exports.start (/double-take/api/src/util/process.util.js:137:28)
at module.exports.polling (/double-take/api/src/util/process.util.js:55:36)
at runMicrotasks ()
at processTicksAndRejections (node:internal/process/task_queues:96:5)
at async Promise.all (index 1)

23-04-24 08:24:57 error: TypeError: aiserver process error: Cannot read properties of undefined (reading ‘recognize’)
at module.exports.recognize (/double-take/api/src/util/detectors/actions/index.js:4:24)
at Object.module.exports.process (/double-take/api/src/util/process.util.js:160:28)
at Object.module.exports.start (/double-take/api/src/util/process.util.js:137:28)
at module.exports.polling (/double-take/api/src/util/process.util.js:55:36)
at runMicrotasks ()
at processTicksAndRejections (node:internal/process/task_queues:96:5)
at async Promise.all (index 0)
at async module.exports.start (/double-take/api/src/controllers/recognize.controller.js:152:7)
23-04-24 08:24:59 info: done processing frontyard: 1682339029.911746-m8kewp in 6.13 sec
23-04-24 08:24:59 info: {
id: ‘1682339029.911746-m8kewp’,
duration: 6.13,
timestamp: ‘2023-04-24T12:24:59.595Z’,
attempts: 19,
camera: ‘frontyard’,
zones: [],
counts: { person: 0, match: 0, miss: 0, unknown: 0 },
matches: [],
misses: [],
unknowns: []

I’ve got doubletake running based on the latest release (v1.13.4). I also have aiserver.io running (2.1.4-Beta (2.1.6 isn’t avail yet)).

I have installed ‘face processing, Object Detection 6.2 and TF-Lite’ as modules.

I have added:

  aiserver:
    url: http://192.168.21.9:32168
    # number of seconds before the request times out and is aborted
    timeout: 15

To the doubleTake config file.

When going to that exact IP incl port with Chrome it shows:

But it times out. Any thoughts?

Is new release double take out yet to update in home assistance or did you manual updated. I still on version 1.13.1.

Greg

I on double take version 1.31.1
I don’t see new version 1.31.4 to update in home assistance.

Greg

Manuel yes. Separately from ha

Can they add new version to home assistance as beta update.

Greg

Same problem as @customstudioyandexru

If you hover your mouse above the red symbol, it shows the last error message.

@Jako being as DeepStack is no more, what would you recommend to use for detector these days?

I built Frigate a week ago on Docker in an LXC in Proxmox and added an M.2 Coral to which brought the inference down to an average of 6ms (reduced from 120ms) so that is working great. Im wondering if there are any detectors that can also utilise the Coral as well?

1 Like

I on double take version 1.31.1
When will version 1.31.4 to ready to update in home assistance. I still getting error use windows 11 codeproject. Is there any thing I need to updated.

I have run ok in blue iris in windows 11.

Thanks

Greg Mack

I getting this error on n double taken. Can someone help
Me resolve this issues please.

info: Double Take v1.13.1-beta.1-b353068
error: 1 validation error
error: detectors is not any of [subschema 0],[subschema 1],[subschema 2],[subschema 3]
info: MQTT: connected
info: MQTT: subscribed to frigate/events, frigate/+/person/snapshot

Try this branch: GitHub - skrashevich/double-take: Unified UI and API for processing and training images for facial recognition.
It has integration with CodeProject. However I am getting issues with everything as a 0% match, but at least you can train from Double Take and get a response from Double Take. If you do use it and get it working with correct match, please do let me know

@ctml did you manage to implement this? Did you use @cunninr2 suggested post? I have tried implementing it but keep getting stuck with a lot of errors on paste

I didn’t figure out to install his beta version in home assistance. It not available for download. If tell me how to do it manually I will try to do it. Right I stuck waiting for new version to be able to be download.

If you have any ideas how we can resolve this issue just let me know and I will try to test it to get it working.

Greg