# Optional: minimum score for the object to initiate tracking (default: 0.5) # Optional: maximum width*height of the bounding box for the detected object (default: 24000000) # Optional: minimum width*height of the bounding box for the detected object (default: 0) If you want different objects being detected, cars for example, make sure you add them too and entities for those will be created. I am assuming you already have an MQTT client installed. My one looks like this for 4 cameras I have changed some passwords for security, your IP addresses for cameras and MQTT client will be different. created a config.yml file in /var/lib/docker/volumes/frigate_config/_data.type: tmpfs # Optional: 1GB of memory, reduces SSD/SD Card wear dev/dri/renderD128 # for intel hwaccel, needs to be updated for your hardware Image: blakeblackshear/frigate:0.9.0-rc3-amd64 Privileged: true # this may not be necessary for all setups My one, for guide looks like this, you will need to change the source to where you want to save the video stream Create a stack in Portainer for Frigate.Remove Portainer from home assistant supervisor if you use it (I had it to run the excellent RAKOMQTT).Restore backup of home assistant and test.Install Home assistant Supervised on Debian.Share USB hard drive connected to the NUC using SAMBA on Cockpit.Install cockpit for easy access to Debian using SSH and Terminal.Backup home assistant and copy the files somewhere safe.Here are my Reolink settings for your guide: The cameras will only wrk on Frigate up to 2.7k not the full 4k, again this may be fixed by the time you read this. This may not be necessary now as they may have changed their firmware. Install the beta firmware on the reolink cameras by contacting reolink support.This meant the the best method for me was to install Frigate not on supervisor, but on Portainer on the same machine that runs home assistant. However an issue with Frigate when it is installed in Home Assistant supervisor is that the video is saved to the media folder and not to a USB or SAMBA drive. The main benefit of this is that I no longer need Windows 10 on the NUC, allowing me to install Home Assistant without using Proxmox, making it much easier to access the Coral accelerator, for example. Under the guidance of Pavol, a very helpful member of the Home Assistant Face Book group, I decided to go down a different path: I probably had my proxmox allocations set up incorrectly. I started to get performance issues in Home Assistant due to the extra resources required by Frigate.I could not get pass through for a USB coral accelerator which I bought to improve Frigate performance.I could not get quality streams recoding on Blue Iris despite many attempts.An alternative was to also use Frigate in Home Assistant for the object recognition and automations and to keep Blue Iris purely for the 24/7 recording. The idea was that Win 10 would run Blue Iris saving streaming video to the hard drive and I would then be able to trigger automations in Home Assistant using MQTT. My hardware set up was am Intel i7 NUC with proxmox running HA OS, Windows 10 and Open Media Vault sharing a large external HDD using Samba share. This was particularly frustrating as so many quality youtubers recommended the cameras and Blue Iris but failed to mention that the two together are not a great combination. As I ranted about in this post I really struggled with my Reolink RLC-820A and Reolink RLC-810A cameras with Blue Iris.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |