Hi All
What are you doing about housekeeping of old images/recordings to stop your disks filling up? I can’t see any reference to it anywhere.
Thanks
Dave
My newer 4k cameras support ONVIF with motion detection events. Although currently with Frigate the CPU usage is not extreme, it does take some power from the wall (maybe 15 watts more) to continuously monitor for motion.
I was thinking it would be nice if there were an option to offload the motion detection to the camera using these ONVIF events.
The custom component is not a replacement for the addon. It connects to the addon and integrates it into homeassistant.
You can set a time limit for records and clips .I set mine to one day
But my streams are not fluent …more stuttering. I have 5pics/second for motion detection enabled but I thought the stream should be like the frames from the cam stream ?
If you do that, you might as well not even use Frigate. Just use Blue Iris or something similar and have Blue Iris record clips based off of those ONVIF events.
ffmpeg.Einfahrt.detect ERROR : [h264 @ 0x1c9cdc0] decode_slice_header error
ffmpeg.Einfahrt.detect ERROR : [h264 @ 0x1c9cdc0] no frame!
ffmpeg.Einfahrt.detect ERROR : Guessed Channel Layout for Input Stream #0.1 : mono
ffmpeg.Einfahrt.detect ERROR : [h264_mmal @ 0x1e6b010] Did not get output frame from MMAL.
ffmpeg.Einfahrt.detect ERROR : Error while decoding stream #0:0: Unknown error occurred
ffmpeg.Einfahrt.detect ERROR : [rtsp @ 0x1c99710] CSeq 12 expected, 0 received.
ffmpeg.Einfahrt.detect ERROR : Last message repeated 18 times
ffmpeg.Einfahrt.detect ERROR : [NULL @ 0x1c9cdc0] missing picture in access unit with size 1440
ffmpeg.Einfahrt.detect ERROR : [rtsp @ 0x1c99710] CSeq 12 expected, 0 received.
ffmpeg.Einfahrt.detect ERROR : Last message repeated 29 times
ffmpeg.Einfahrt.detect ERROR : [flv @ 0x1cb4d70] Failed to update header with correct duration.
ffmpeg.Einfahrt.detect ERROR : [flv @ 0x1cb4d70] Failed to update header with correct filesize.
[h264 @ 0x10a90b0] out of range intra chroma pred mode
[h264 @ 0x10a90b0] error while decoding MB 10 52
frigate.events WARNING : Unable to create clip for Einfahrt and event 1614524505.768079-kj4a62. There were no cache files for this event.
frigate.events WARNING : Unable to create clip for Einfahrt and event 1614524516.454285-9ymn4n. There were no cache files for this event.
frigate.video INFO : Einfahrt: ffmpeg sent a broken frame. memoryview assignment: lvalue and rvalue have different structures
frigate.video INFO : Einfahrt: ffmpeg sent a broken frame. memoryview assignment: lvalue and rvalue have different structures
frigate.video INFO : Einfahrt: ffmpeg process is not running. exiting capture thread...
[h264 @ 0x9660b0] corrupted macroblock 96 24 (total_coeff=-1)
[h264 @ 0x9660b0] error while decoding MB 96 24
ffmpeg.Einfahrt.detect ERROR : [h264 @ 0x95adc0] non-existing PPS 0 referenced
ffmpeg.Einfahrt.detect ERROR : Last message repeated 1 times
ffmpeg.Einfahrt.detect ERROR : [h264 @ 0x95adc0] decode_slice_header error
ffmpeg.Einfahrt.detect ERROR : [h264 @ 0x95adc0] no frame!
ffmpeg.Einfahrt.detect ERROR : [h264 @ 0x95adc0] non-existing PPS 0 referenced
ffmpeg.Einfahrt.detect ERROR : Last message repeated 1 times
ffmpeg.Einfahrt.detect ERROR : [h264 @ 0x95adc0] decode_slice_header error
ffmpeg.Einfahrt.detect ERROR : [h264 @ 0x95adc0] no frame!
ffmpeg.Einfahrt.detect ERROR : [h264 @ 0x95adc0] non-existing PPS 0 referenced
ffmpeg.Einfahrt.detect ERROR : Last message repeated 1 times
ffmpeg.Einfahrt.detect ERROR : [h264 @ 0x95adc0] decode_slice_header error
ffmpeg.Einfahrt.detect ERROR : [h264 @ 0x95adc0] no frame!
ffmpeg.Einfahrt.detect ERROR : [h264 @ 0x95adc0] non-existing PPS 0 referenced
ffmpeg.Einfahrt.detect ERROR : Last message repeated 1 times
ffmpeg.Einfahrt.detect ERROR : [h264 @ 0x95adc0] decode_slice_header error
ffmpeg.Einfahrt.detect ERROR : [h264 @ 0x95adc0] no frame!
ffmpeg.Einfahrt.detect ERROR : Guessed Channel Layout for Input Stream #0.1 : mono
ffmpeg.Einfahrt.detect ERROR : [rtsp @ 0x957710] CSeq 11 expected, 0 received.
ffmpeg.Einfahrt.detect ERROR : Last message repeated 58 times
ffmpeg.Einfahrt.detect ERROR : [flv @ 0x98b9c0] Failed to update header with correct duration.
ffmpeg.Einfahrt.detect ERROR : [flv @ 0x98b9c0] Failed to update header with correct filesize.
Does anybody know this kind of error? It’s only happening with my Foscam camera. The other cameras are working fine.
Thank you.
This is one of those “different strokes for different folks” things.
I personally don’t see Frigate as a replacement for a traditional NVR as it just doesn’t have the same features.
Therefore I do run Blue Iris still. I really wish it worked on linux, but there isn’t anything that matches the feature set for the price so I still have a windows box running just for that.
I use Frigate for notifications of people or vehicles in my driveway and that is all I use it for.
It works better than tensorflow or deepstack…much fewer false positives than either one of them.
I use mine the same way. Frigate for object detection/automation triggering and Synology for long term recording.
I’m not sure you understand what I’m saying. I would replace Frigate’s motion detection with the camera motion detection. Frigate would still track and detect the objects.
I too still run BI for 24x7 recording and use Frigate just for notifications for my phone. Both work really well.
Ok thanks, I was also hoping to be able to use the 2 streams from my camera one as the frigate detection running at 640x480 5fps and when triggers record the main stream which is 1280x960 25fps but that’s not a feature…yet…hopefully
Hey guys,
How do you setup push notifications for frigate events?
I use binary template sensors to determine if someone is detected when the alarm is armed (i.e. no one is home or we’re gone to bed).
intruder_driveway:
unique_id: intruder_driveway
friendly_name: Intruder driveway
value_template: >
{{ is_state("binary_sensor.driveway_person","on") and
is_state("binary_sensor.alarm_armed","on") }}
Then I use Telegram to send a snapshot of the detected person. Sending the snapshot is done with a script and triggering that script is done through an automation triggered by the previous binary sensor.
alias: Alert bot - Photo - Driveway person
sequence:
- service: notify.alert_bot
data:
message: Person detected near driveway
data:
photo:
caption: '{{ now() | as_timestamp | timestamp_custom("%y-%m-%d %H:%M:%S") }}'
url: >-
http://hass.local:8123{{
state_attr("camera.driveway_person","entity_picture") }}
mode: single
and
alias: Alert bot - Driveway intruder detected
description: ''
trigger:
- platform: state
entity_id: binary_sensor.intruder_driveway
condition: []
action:
- service: script.alert_bot_driveway_person_image
data: {}
mode: single
I have the Telegram integration setup three times. Each sends a message to a different group (info, warning, or alert). Each group has different notification settings (silent, pop-up, or pop-up and ringer). Frigate events go to the alert group.
I repeat this same setup for a second camera as well.
1 Like
I use this automation:
- alias: Notify of frigate person events
trigger:
platform: mqtt
topic: frigate/events
condition:
- "{{ ( as_timestamp(now()) - as_timestamp(state_attr('automation.notify_of_frigate_person_events', 'last_triggered')) |int(0) ) > 30 }}"
- "{{ trigger.payload_json['after']['label'] == 'person' }}"
- condition: state
entity_id: alarm_control_panel.alarm
state: armed_away
action:
- service: notify.all_phones
data_template:
message: 'A {{trigger.payload_json["after"]["label"]}} was detected by the {{trigger.payload_json["after"]["camera"]}} camera.'
data:
image: 'https://xxx.xxx.xxx.xxx/api/frigate/notifications/{{trigger.payload_json["after"]["id"]}}/thumbnail.jpg?format=android'
tag: '{{trigger.payload_json["after"]["id"]}}'
It sends an HTML5 notification if a person is detected on any of my cameras when my alarm is armed away, along with an image.
Frigate works perfect if the whole person is visible.
Is it possible to detect a person behind a solid fence?
Something like haarcascade upperbody detection…
Does anyone have experience with this?
1 Like
Hi,
i’m trying to run frigate on a dedicated host (without HA)
this is my config.yml
mqtt:
host: 192.168.1.100:81
topic_prefix: frigate
user: mqttuser
password: mqttpassword
frigate.mqtt ERROR: Unable to connect to MQTT server: [Errno -2] Name or service not know
any ideas?
I’ve had Frigate detect people showing only their upper body on our TV before.
I have a question regarding sending snapshots. I have two zones set up and I have mqtt set up to only send when an object enters the zones. I would normally grab the snapshot from the camera object entity in HA like this:
trigger:
platform: state
entity_id: binary_sensor.front_drive_person_motion
to: 'on'
action:
- service: notify.mobile_app_my_phone
data_template:
title: "MOTION ALERT"
message: "Front Camera - Person detected on the drive"
data:
image: "/api/camera_proxy/camera.front_drive_person"
But I’m not getting any MQTT messages sent for topic frigate/front/person so the camera in HA is blank. Also, frigate/front/car has stopped showing in HA for some reason the last image being two hours ago yet I’ve got several snapshots in MQTT explorer since. Other cameras are updating fine.
nevermind, i found the error
host: 192.168.1.100:81
have to be
host: 192.168.1.100
port: 81