I am not able to make a request with the deepstack_object or deepstack_face component. I just get ConnectionError: Is deepstack_object running?. Has anyone else had this issue. It isn’t coming through on the server either and is just hanging on the HASS side. Port and IP Address are defiantly correct.
Did you add the activation key?
what version are you using (CPU / GPU)?
Not yet, I’m using the component with the adjustments I added to the HA component, running face recognitation every 2 seconds, my problem is currently not the memory leak, more like the fact that I can’t run multiple requests at the same timeframe of 1 second, which is being investigated by deepstack
Yes I have activated the api on the server. Before I realised that I needed to activate it, I used to get another error but it used to actually come through in the terminal of where I was running Deepstack, now, nothing comes through. I am running on a docker container on an old Mac so I’m sure it’s CPU.
I worked it out. Turns out that my old Mac from 2011 didn’t support avx so I had to install the noavx version with sudo docker pull deepquestai/deepstack:noavx and it now works perfectly. Thanks for the great component!
2 Likes
Any tricks to registering faces? No matter what I try i get this error. If I put the url in the browser it finds it no issues.
Log Details (ERROR)
deepstack_face error: Invalid file path: http://192.168..:8123/local/images/Alexi1.jpg
It works with local storage not http protocol, let me know if it solves the problem
Ok I tried with local, and now there is no error, but it also doesnt seem like anything changed (not recognising faces). Should there be a message or entry in the log if the face upload was successful?
Elad hi,
I’m using your great component but having some troubles.
i’m getting this error:
“TypeError: detect() missing 1 required positional argument: ‘targets’”
but the “trgets” section in the config file is like in your sample.
What am i missing?
Thanks
Can you provide more details? In which stage does it happen? Configuration?
Thanks
There’s a aervice to list all registered faces, it will create a persistent notification (HA notification) with all faces:
deepstack.list_faces
It’s hapaning when i’m calling the service or by the interval.
if i’m using the configuration checker everything is ok.
Thanks,
Has there been any movement on deepstacks event firing yet? I would love to try and develop something but I have not heard anything as of late
@lootiejay To add this feature you can use the Facebox component as a guide. I am happy to review/accept PR on the Deepstack CC but am to busy to work on new features myself
when using `
sudo docker run -v localstorage:/datastore -p 5000:5000 deepquestai/deepstack
` getting that error
standard_init_linux.go:207: exec user process caused "exec format error"
Hello Guys,
Just wondering if anyone got deepstack component to work with A ring camera. I am unable to teach a face. I run my instance on Ubuntu 18.02 in docker and when I use image_processing_teach_face nothing happens and do not even get a message in my logs
Thanks in advance
looking awsome, could yuo add support for custom_updater: so it’s easy to see whrn there is an update? thanks 
Or perhaps switching to something else like HACS, i think custom_updater won’t get many updates.
Hi all
I just released v0.7 which adds an event that is fired when objects are detected, allowing notifications to be fired when target objects are detected. I still need to figure out how to include the camera image in notifications, perhaps someone in the community already knows
Event image_processing.object_detected
An event image_processing.object_detected is fired for each object detected. The event payload includes:
-
classifier: the classifier (i.e.deepstack_object) -
entity_id: the entity id responsible for the event -
object: the object detected -
confidence: the confidence in detection in the range 0 - 1 where 1 is 100% confidence.
An example automation using the event is given below:
- action:
- data_template:
title: "New object detection"
message: "{{ trigger.event.data.object }} with confidence {{ trigger.event.data.confidence }}"
service: notify.pushbullet
alias: Object detection automation
condition: []
id: '1120092824622'
trigger:
- platform: event
event_type: image_processing.object_detected
event_data:
object: person
1 Like
Hi all
just published releasev0.8 which adds saving of processed images with bounding boxes around target objects. This makes it really easy to include the bounding box image in notifications like below.
Cheers
1 Like
is this supposed to work with rtsp stream camera entities ?
Iam somehow unsure about
source:
- entity_id: camera.local_file
name: face_counter
I put also my amcrest camera but nothing is detetected at all. Either ffmpeg rtsp or amcrest platform
deppstack is running and can be accessed via web ifc or API. I sucessfully registered faces. There is no log entry in home assistant.
Any ideas =?
PS: in deepstack I dont see any incoming requests from HA
Yes your config is wrong, and you have not posted it here