yayitazale

Of course 4K requires much more power to be processed...

Face Recognition | FrigateFace recognition identifies known individuals by matching detected faces with previously learned facial data. When a known person is recognized, their name will be added as a sub_label. This informati

Width and height missing... Copy the proposed configuration

You did not copy the required input_dtype config https://docs.frigate.video/configuration/object_detectors#yolo-v3-v4-v7-v9-1

I just set up yolov9 with OpenVino and I see also "INFO: Created TensorFlow Lite XNNPACK delegate for CPU." but the system is using the openvino GPU correctly. Waht do you see in the /system#general As you can see in the images below, in my case as I'm using openvino I see a OV as detector. This are my stats with a T size 320 model: With a T size 640 model: With a M size 320 model: With a M size 640 model: If I switch back to Coral, I see coral as detector:

No, in the host. Those are not yolo models -> https://docs.frigate.video/configuration/object_detectors/#downloading-yolo-models

Yes, they are already removed from the master template but will no make any difference to you until you reinstall the app, so the easiest way is just remove the entries.

From what I understand after reading the documentation, the generation of YOLO models for TensorRT is now only available for Jetson Nano. In your case, you have to switch to use ONNX models, as explained here. The usage of the tensorRT version is still needed, but you have to manually download and place the model.

And what do you see in the logs?

Did you added "--runtime=nvidia" as extra parameter under advanced view?

Tensor version is supported in version 16 https://github.com/blakeblackshear/frigate/releases/tag/v0.16.0. Models will be created automatically with tensorrt version if you have correctly defined YOLO_MODELS varibale and the rest of the template https://docs.frigate.video/configuration/object_detectors#generate-models

I don't know how you are launching the container but it is already prepared for yolo models. Are you using the tensor branch? did you have the nvidia driver installed? https://docs.frigate.video/configuration/object_detectors#generate-models

@ich777 I don't know if you can help us with this.

You don't need to use "--gpus all" as the GPU pass-through of a specific GPU to the container is done with the "nvidia visible devices" and "nvidia driver capabilities" variable. If you are using the GPU only for decoding/encoding then you can delete the YOLO_MODELS variable as this is only use for GPU detection. What you can see about the GPU decoding in the logs of frigate? Is go2rtc working correctly?

I don't know what issue can you have then.

I guess your Docker image is not big enough for the tensorrt image. Go to Settings > Docker, then disable Docker and increase the Docker Image Size to at least 8GB. This tensort version of frigate requires at least 4-5GB.

You can participate in the development https://github.com/blakeblackshear/frigate There is no "development" in unraid, I just created and maintain the template to run this app in unraid.

Port 5000 is a legacy port and will be only used for API in the future. I'm not part of the developers team of the project but I assume that the connection to "huggingface.co" is done to donwload semantic search models https://docs.frigate.video/configuration/semantic_search.

Edit the container, click on advanced view (top right) and then you have to edit extra parameters: This is my config for 6 full hd cams: --shm-size=562m --mount type=tmpfs,target=/tmp/cache,tmpfs-size=1000000000 --restart unless-stopped For SHM size calculation: https://docs.frigate.video/frigate/installation#calculating-required-shm-size For tmpfs there is no guide, I'm currently using 1GB with no issues.

If you don't plan to use this rtsp streams anywhere else, then I think there is no problem to delete them. https://docs.frigate.video/configuration/authentication#onboarding With a Admin user You can try https://www.advanced-ip-scanner.com/es/

I changed the server a few months ago to an Intel Pentium Gold 8505 - 32 GB DDR5 - Integrated Intel i915 Alder Lake-UP3 GT1 GPU. Previously I had an Intel Pentium Gold G5600 - 32 Gb DDR4 - Nvidia Gforce GTX1060 3GB dedicated GPU. With the old server and all cameras running I had about 25-30% CPU usage, and now with the new server I barely see more than 20%. I usually have 13-14% CPU usage with all cameras on during the night, and during the day with 4 cameras I have about 8% CPU usage, with Frigate always being the biggest consumer. Edit: I use the GPU for FFMPEG restreaming.

I have never used Motioneye so I don't know if the CPU usage will be higher or lower. About USB Coral, I have been using one for the last 3 years non-stop with frigate (8 cameras FHD) and it works perfect.

Thanks for the update.

Seems that your detection is not working Did you have correct drivers installend in using the nvidia plugin? Are you using the nvidia tensorRT branch in the installation? Did you added "--runtime=nvidia" as extra parameter? Did you correctly passtrought the new GPU? This part doesn't match with documentation: detectors: tensorrt: type: tensorrt device: 0 model: path: /config/model_cache/tensorrt/8.5.3/yolov7-tiny-416.trt input_tensor: nchw input_shape: [1, 3, 416, 416] https://docs.frigate.video/configuration/object_detectors#configuration-parameters You should change input shape to width: 416 height: 416

Default in the template: --shm-size=256m --mount type=tmpfs,target=/tmp/cache,tmpfs-size=1000000000 --restart unless-stopped