Skip to main content

Supported Hardware

info

Frigate supports multiple different detectors that work on different types of hardware:

Most Hardware

  • Coral EdgeTPU: The Google Coral EdgeTPU is available in USB, Mini PCIe, and m.2 formats allowing for a wide range of compatibility with devices.
  • Hailo: The Hailo8 and Hailo8L AI Acceleration module is available in m.2 format with a HAT for RPi devices, offering a wide range of compatibility with devices.
  • Community Supported MemryX: The MX3 Acceleration module is available in m.2 format, offering broad compatibility across various platforms.
  • Community Supported DeGirum: Service for using hardware devices in the cloud or locally. Hardware and models provided on the cloud on their website.

AMD

  • ROCm: ROCm can run on AMD Discrete GPUs to provide efficient object detection.
  • ONNX: ROCm will automatically be detected and used as a detector in the -rocm Frigate image when a supported ONNX model is configured.

Apple Silicon

  • Apple Silicon: Apple Silicon can run on M1 and newer Apple Silicon devices.

Intel

  • OpenVino: OpenVino can run on Intel Arc GPUs, Intel integrated GPUs, and Intel CPUs to provide efficient object detection.
  • ONNX: OpenVINO will automatically be detected and used as a detector in the default Frigate image when a supported ONNX model is configured.

Nvidia GPU

  • ONNX: Nvidia GPUs will automatically be detected and used as a detector in the -tensorrt Frigate image when a supported ONNX model is configured.

Nvidia Jetson Community Supported

  • TensortRT: TensorRT can run on Jetson devices, using one of many default models.
  • ONNX: TensorRT will automatically be detected and used as a detector in the -tensorrt-jp6 Frigate image when a supported ONNX model is configured.

Rockchip Community Supported

  • RKNN: RKNN models can run on Rockchip devices with included NPUs.

Synaptics Community Supported

  • Synaptics: synap models can run on Synaptics devices(e.g astra machina) with included NPUs.

AXERA Community Supported

  • AXEngine: axmodels can run on AXERA AI acceleration.

For Testing

note

Multiple detectors can not be mixed for object detection (ex: OpenVINO and Coral EdgeTPU can not be used for object detection at the same time).

This does not affect using hardware for accelerating other tasks such as semantic search

Officially Supported Detectors

Frigate provides a number of builtin detector types. By default, Frigate will use a single OpenVINO detector running on the CPU. Other detectors may require additional configuration as described below. When using multiple detectors they will run in dedicated processes, but pull from a common queue of detection requests from across all cameras.

Edge TPU Detector​

The Edge TPU detector type runs TensorFlow Lite models utilizing the Google Coral delegate for hardware acceleration. To configure an Edge TPU detector, set the "type" attribute to "edgetpu".

The Edge TPU device can be specified using the "device" attribute according to the Documentation for the TensorFlow Lite Python API. If not set, the delegate will use the first device it finds.

tip

See common Edge TPU troubleshooting steps if the Edge TPU is not detected.

Single USB Coral​

Navigate to Settings→System→Detector hardware and select EdgeTPU from the detector type dropdown and click Add, then set device to usb.

Multiple USB Corals​

Navigate to Settings→System→Detector hardware and select EdgeTPU from the detector type dropdown and click Add to add multiple detectors, specifying usb:0 and usb:1 as the device for each.

Native Coral (Dev Board)​

warning: may have compatibility issues after v0.9.x

Navigate to Settings→System→Detector hardware and select EdgeTPU from the detector type dropdown and click Add, then leave the device field empty.

Single PCIE/M.2 Coral​

Navigate to Settings→System→Detector hardware and select EdgeTPU from the detector type dropdown and click Add, then set device to pci.

Multiple PCIE/M.2 Corals​

Navigate to Settings→System→Detector hardware and select EdgeTPU from the detector type dropdown and click Add to add multiple detectors, specifying pci:0 and pci:1 as the device for each.

Mixing Corals​

Navigate to Settings→System→Detector hardware and select EdgeTPU from the detector type dropdown and click Add to add multiple detectors with different device types (e.g., usb and pci).

EdgeTPU Supported Models​

ModelNotes
MobiledetDefault model
YOLOv9More accurate but slower than default model

Mobiledet​

A TensorFlow Lite model is provided in the container at /edgetpu_model.tflite and is used by this detector type by default. To provide your own model, bind mount the file into the container and provide the path with model.path.

YOLOv9​

YOLOv9 models that are compiled for TensorFlow Lite and properly quantized are supported, but not included by default. Instructions for downloading a model with support for the Google Coral.

tip

Frigate+ Users: Follow the instructions to set a model ID in your config file.

YOLOv9 Setup & Config

After placing the downloaded files for the tflite model and labels in your config folder, use the following configuration:

Navigate to Settings→System→Detector hardware and select EdgeTPU from the detector type dropdown and click Add, then set device to usb. Then navigate to Settings→System→Detection model and configure the model settings:

FieldValue
Object Detection Model Typeyolo-generic
Object detection model input width320 (should match the imgsize of the model)
Object detection model input height320 (should match the imgsize of the model)
Custom object detector model path/config/model_cache/yolov9-s-relu6-best_320_int8_edgetpu.tflite
Label map for custom object detector/config/labels-coco17.txt

Note that due to hardware limitations of the Coral, the labelmap is a subset of the COCO labels and includes only 17 object classes.

Hailo-8​

This detector is available for use with both Hailo-8 and Hailo-8L AI Acceleration Modules. The integration automatically detects your hardware architecture via the Hailo CLI and selects the appropriate default model if no custom model is specified.

See the installation docs for information on configuring the Hailo hardware.

info

If no custom model is provided, the Hailo detector downloads a default model from the Hailo Model Zoo on first startup. Once cached, the model works fully offline. See Network Requirements for details.

Configuration​

When configuring the Hailo detector, you have two options to specify the model: a local path or a URL. If both are provided, the detector will first check for the model at the given local path. If the file is not found, it will download the model from the specified URL. The model file is cached under /config/model_cache/hailo.

YOLO​

Use this configuration for YOLO-based models. When no custom model path or URL is provided, the detector automatically downloads the default model based on the detected hardware:

  • Hailo-8 hardware: Uses YOLOv6n (default: yolov6n.hef)
  • Hailo-8L hardware: Uses YOLOv6n (default: yolov6n.hef)

Navigate to Settings→System→Detector hardware and select Hailo-8/Hailo-8L from the detector type dropdown and click Add, then set device to PCIe. Then navigate to Settings→System→Detection model and configure the model settings:

FieldValue
Object detection model input width320
Object detection model input height320
Model Input Tensor Shapenhwc
Model Input Pixel Color Formatrgb
Model Input D Typeint
Object Detection Model Typeyolo-generic
Label map for custom object detector/labelmap/coco-80.txt

The detector automatically selects the default model based on your hardware. Optionally, specify a local model path or URL to override.

SSD​

For SSD-based models, provide either a model path or URL to your compiled SSD model. The integration will first check the local path before downloading if necessary.

Navigate to Settings→System→Detector hardware and select Hailo-8/Hailo-8L from the detector type dropdown and click Add, then set device to PCIe. Then navigate to Settings→System→Detection model and configure the model settings:

FieldValue
Object detection model input width300
Object detection model input height300
Model Input Tensor Shapenhwc
Model Input Pixel Color Formatrgb
Object Detection Model Typessd

Specify the local model path or URL for SSD MobileNet v1.

Custom Models​

The Hailo detector supports all YOLO models compiled for Hailo hardware that include post-processing. You can specify a custom URL or a local path to download or use your model directly. If both are provided, the detector checks the local path first.

Navigate to Settings→System→Detector hardware and select Hailo-8/Hailo-8L from the detector type dropdown and click Add, then set device to PCIe. Then navigate to Settings→System→Detection model and configure the model settings to match your custom model dimensions and format.

For additional ready-to-use models, please visit: https://github.com/hailo-ai/hailo_model_zoo

Hailo8 supports all models in the Hailo Model Zoo that include HailoRT post-processing. You're welcome to choose any of these pre-configured models for your implementation.

Note: The config.path parameter can accept either a local file path or a URL ending with .hef. When provided, the detector will first check if the path is a local file path. If the file exists locally, it will use it directly. If the file is not found locally or if a URL was provided, it will attempt to download the model from the specified URL.

OpenVINO Detector​

The OpenVINO detector type runs an OpenVINO IR model on AMD and Intel CPUs, Intel GPUs and Intel NPUs. To configure an OpenVINO detector, set the "type" attribute to "openvino".

The OpenVINO device to be used is specified using the "device" attribute according to the naming conventions in the Device Documentation. The most common devices are CPU, GPU, or NPU.

OpenVINO is supported on 6th Gen Intel platforms (Skylake) and newer. It will also run on AMD CPUs despite having no official support for it. A supported Intel platform is required to use the GPU or NPU device with OpenVINO. For detailed system requirements, see OpenVINO System Requirements

tip

NPU + GPU Systems: If you have both NPU and GPU available (Intel Core Ultra processors), use NPU for object detection and GPU for enrichments (semantic search, face recognition, etc.) for best performance and compatibility.

When using many cameras one detector may not be enough to keep up. Multiple detectors can be defined assuming GPU resources are available. An example configuration would be:

Navigate to Settings→System→Detector hardware and select OpenVINO from the detector type dropdown and click Add to add multiple detectors, each targeting GPU or NPU.

OpenVINO Supported Models​

ModelGPUNPUNotes
YOLOv9✅✅Recommended for GPU & NPU
RF-DETR✅✅Requires XE iGPU or Arc
YOLO-NAS✅✅
MobileNet v2✅✅Fast and lightweight model, less accurate than larger models
YOLOX✅?
D-FINE / DEIMv2❌❌

SSDLite MobileNet v2​

An OpenVINO model is provided in the container at /openvino-model/ssdlite_mobilenet_v2.xml and is used by this detector type by default. The model comes from Intel's Open Model Zoo SSDLite MobileNet V2 and is converted to an FP16 precision IR model.

MobileNet v2 Config

Use the model configuration shown below when using the OpenVINO detector with the default OpenVINO model:

Navigate to Settings→System→Detector hardware and select OpenVINO from the detector type dropdown and click Add, then set device to GPU (or NPU). Then navigate to Settings→System→Detection model and configure:

FieldValue
Object detection model input width300
Object detection model input height300
Model Input Tensor Shapenhwc
Model Input Pixel Color Formatbgr
Custom object detector model path/openvino-model/ssdlite_mobilenet_v2.xml
Label map for custom object detector/openvino-model/coco_91cl_bkgr.txt

YOLOX​

This detector also supports YOLOX. Frigate does not come with any YOLOX models preloaded, so you will need to supply your own models.

YOLO-NAS​

YOLO-NAS models are supported, but not included by default. See the models section for more information on downloading the YOLO-NAS model for use in Frigate.

YOLO-NAS Setup & Config

After placing the downloaded onnx model in your config folder, use the following configuration:

Navigate to Settings→System→Detector hardware and select OpenVINO from the detector type dropdown and click Add, then set device to GPU. Then navigate to Settings→System→Detection model and configure:

FieldValue
Object Detection Model Typeyolonas
Object detection model input width320 (should match whatever was set in notebook)
Object detection model input height320 (should match whatever was set in notebook)
Model Input Tensor Shapenchw
Model Input Pixel Color Formatbgr
Custom object detector model path/config/yolo_nas_s.onnx
Label map for custom object detector/labelmap/coco-80.txt

Note that the labelmap uses a subset of the complete COCO label set that has only 80 objects.

YOLO (v3, v4, v7, v9)​

YOLOv3, YOLOv4, YOLOv7, and YOLOv9 models are supported, but not included by default.

tip

The YOLO detector has been designed to support YOLOv3, YOLOv4, YOLOv7, and YOLOv9 models, but may support other YOLO model architectures as well.

YOLOv Setup & Config
warning

If you are using a Frigate+ model, you should not define any of the below model parameters in your config except for path. See the Frigate+ model docs for more information on setting up your model.

After placing the downloaded onnx model in your config folder, use the following configuration:

Navigate to Settings→System→Detector hardware and select OpenVINO from the detector type dropdown and click Add, then set device to GPU (or NPU). Then navigate to Settings→System→Detection model and configure:

FieldValue
Object Detection Model Typeyolo-generic
Object detection model input width320 (should match the imgsize set during model export)
Object detection model input height320 (should match the imgsize set during model export)
Model Input Tensor Shapenchw
Model Input D Typefloat
Custom object detector model path/config/model_cache/yolo.onnx
Label map for custom object detector/labelmap/coco-80.txt

Note that the labelmap uses a subset of the complete COCO label set that has only 80 objects.

RF-DETR​

RF-DETR is a DETR based model. The ONNX exported models are supported, but not included by default. See the models section for more informatoin on downloading the RF-DETR model for use in Frigate.

warning

Due to the size and complexity of the RF-DETR model, it is only recommended to be run with discrete Arc Graphics Cards.

RF-DETR Setup & Config

After placing the downloaded onnx model in your config/model_cache folder, use the following configuration:

Navigate to Settings→System→Detector hardware and select OpenVINO from the detector type dropdown and click Add, then set device to GPU. Then navigate to Settings→System→Detection model and configure:

FieldValue
Object Detection Model Typerfdetr
Object detection model input width320
Object detection model input height320
Model Input Tensor Shapenchw
Model Input D Typefloat
Custom object detector model path/config/model_cache/rfdetr.onnx

D-FINE / DEIMv2​

D-FINE and DEIMv2 are DETR based models that share the same ONNX input/output format. The ONNX exported models are supported, but not included by default. See the models section for downloading D-FINE or DEIMv2 for use in Frigate.

warning

Currently D-FINE / DEIMv2 models only run on OpenVINO in CPU mode, GPUs currently fail to compile the model

D-FINE Setup & Config

After placing the downloaded onnx model in your config/model_cache folder, use the following configuration:

Navigate to Settings→System→Detector hardware and select OpenVINO from the detector type dropdown and click Add, then set device to CPU. Then navigate to Settings→System→Detection model and configure:

FieldValue
Object Detection Model Typedfine
Object detection model input width640
Object detection model input height640
Model Input Tensor Shapenchw
Model Input D Typefloat
Custom object detector model path/config/model_cache/dfine-s.onnx
Label map for custom object detector/labelmap/coco-80.txt

Note that the labelmap uses a subset of the complete COCO label set that has only 80 objects.

DEIMv2 Setup & Config

After placing the downloaded onnx model in your config/model_cache folder, you can use the following configuration:

detectors:
  ov:
    type: openvino
    device: CPU

model:
  model_type: dfine
  width: 640
  height: 640
  input_tensor: nchw
  input_dtype: float
  path: /config/model_cache/deimv2_hgnetv2_n.onnx
  labelmap_path: /labelmap/coco-80.txt

Note that the labelmap uses a subset of the complete COCO label set that has only 80 objects.

Apple Silicon detector​

The NPU in Apple Silicon can't be accessed from within a container, so the Apple Silicon detector client must first be setup. It is recommended to use the Frigate docker image with -standard-arm64 suffix, for example ghcr.io/blakeblackshear/frigate:stable-standard-arm64.

Setup​

  1. Setup the Apple Silicon detector client and run the client
  2. Configure the detector in Frigate and startup Frigate

Configuration​

Using the detector config below will connect to the client:

Navigate to Settings→System→Detector hardware and select ZMQ IPC from the detector type dropdown and click Add, then set the endpoint to tcp://host.docker.internal:5555.

Apple Silicon Supported Models​

There is no default model provided, the following formats are supported:

YOLO (v3, v4, v7, v9)​

YOLOv3, YOLOv4, YOLOv7, and YOLOv9 models are supported, but not included by default.

tip

The YOLO detector has been designed to support YOLOv3, YOLOv4, YOLOv7, and YOLOv9 models, but may support other YOLO model architectures as well. See the models section for more information on downloading YOLO models for use in Frigate.

When Frigate is started with the following config it will connect to the detector client and transfer the model automatically:

Navigate to Settings→System→Detector hardware and select ZMQ IPC from the detector type dropdown and click Add, then set the endpoint to tcp://host.docker.internal:5555. Then navigate to Settings→System→Detection model and configure:

FieldValue
Object Detection Model Typeyolo-generic
Object detection model input width320 (should match the imgsize set during model export)
Object detection model input height320 (should match the imgsize set during model export)
Model Input Tensor Shapenchw
Model Input D Typefloat
Custom object detector model path/config/model_cache/yolo.onnx
Label map for custom object detector/labelmap/coco-80.txt

Note that the labelmap uses a subset of the complete COCO label set that has only 80 objects.

AMD/ROCm GPU detector​

Setup​

Support for AMD GPUs is provided using the ONNX detector. In order to utilize the AMD GPU for object detection use a frigate docker image with -rocm suffix, for example ghcr.io/blakeblackshear/frigate:stable-rocm.

Docker settings for GPU access​

ROCm needs access to the /dev/kfd and /dev/dri devices. When docker or frigate is not run under root then also video (and possibly render and ssl/_ssl) groups should be added.

When running docker directly the following flags should be added for device access:

$ docker run --device=/dev/kfd --device=/dev/dri  \
    ...

When using Docker Compose:

services:
  frigate:
    ...
    devices:
      - /dev/dri
      - /dev/kfd

For reference on recommended settings see running ROCm/pytorch in Docker.

Docker settings for overriding the GPU chipset​

Your GPU might work just fine without any special configuration but in many cases they need manual settings. AMD/ROCm software stack comes with a limited set of GPU drivers and for newer or missing models you will have to override the chipset version to an older/generic version to get things working.

Also AMD/ROCm does not "officially" support integrated GPUs. It still does work with most of them just fine but requires special settings. One has to configure the HSA_OVERRIDE_GFX_VERSION environment variable. See the ROCm bug report for context and examples.

For the rocm frigate build there is some automatic detection:

  • gfx1031 -> 10.3.0
  • gfx1103 -> 11.0.0

If you have something else you might need to override the HSA_OVERRIDE_GFX_VERSION at Docker launch. Suppose the version you want is 10.0.0, then you should configure it from command line as:

$ docker run -e HSA_OVERRIDE_GFX_VERSION=10.0.0 \
    ...

When using Docker Compose:

services:
  frigate:
    ...
    environment:
      HSA_OVERRIDE_GFX_VERSION: "10.0.0"

Figuring out what version you need can be complicated as you can't tell the chipset name and driver from the AMD brand name.

  • first make sure that rocm environment is running properly by running /opt/rocm/bin/rocminfo in the frigate container -- it should list both the CPU and the GPU with their properties
  • find the chipset version you have (gfxNNN) from the output of the rocminfo (see below)
  • use a search engine to query what HSA_OVERRIDE_GFX_VERSION you need for the given gfx name ("gfxNNN ROCm HSA_OVERRIDE_GFX_VERSION")
  • override the HSA_OVERRIDE_GFX_VERSION with relevant value
  • if things are not working check the frigate docker logs

Figuring out if AMD/ROCm is working and found your GPU​

$ docker exec -it frigate /opt/rocm/bin/rocminfo

Figuring out your AMD GPU chipset version:​

We unset the HSA_OVERRIDE_GFX_VERSION to prevent an existing override from messing up the result:

$ docker exec -it frigate /bin/bash -c '(unset HSA_OVERRIDE_GFX_VERSION && /opt/rocm/bin/rocminfo |grep gfx)'

ROCm Supported Models​

tip

The AMD GPU kernel is known problematic especially when converting models to mxr format. The recommended approach is:

  1. Disable object detection in the config.
  2. Startup Frigate with the onnx detector configured, the main object detection model will be converted to mxr format and cached in the config directory.
  3. Once this is finished as indicated by the logs, enable object detection in the UI and confirm that it is working correctly.
  4. Re-enable object detection in the config.

See ONNX supported models for supported models, there are some caveats:

  • D-FINE / DEIMv2 models are not supported
  • YOLO-NAS models are known to not run well on integrated GPUs

ONNX​

ONNX is an open format for building machine learning models, Frigate supports running ONNX models on CPU, OpenVINO, ROCm, and TensorRT. On startup Frigate will automatically try to use a GPU if one is available.

info

If the correct build is used for your GPU then the GPU will be detected and used automatically.

  • AMD

    • ROCm will automatically be detected and used with the ONNX detector in the -rocm Frigate image.

  • Intel

    • OpenVINO will automatically be detected and used with the ONNX detector in the default Frigate image.

  • Nvidia

    • Nvidia GPUs will automatically be detected and used with the ONNX detector in the -tensorrt Frigate image.
    • Jetson devices will automatically be detected and used with the ONNX detector in the -tensorrt-jp6 Frigate image.
tip

When using many cameras one detector may not be enough to keep up. Multiple detectors can be defined assuming GPU resources are available. An example configuration would be:

Navigate to Settings→System→Detector hardware and select ONNX from the detector type dropdown and click Add to add multiple detectors.

ONNX Supported Models​

ModelNvidia GPUAMD GPUNotes
YOLOv9✅✅Supports CUDA Graphs for optimal Nvidia performance
RF-DETR✅⚠ī¸Supports CUDA Graphs for optimal Nvidia performance
YOLO-NAS⚠ī¸âš ī¸Not supported by CUDA Graphs
YOLOX✅✅Supports CUDA Graphs for optimal Nvidia performance
D-FINE / DEIMv2⚠ī¸âŒNot supported by CUDA Graphs

There is no default model provided, the following formats are supported:

YOLO-NAS​

YOLO-NAS models are supported, but not included by default. See the models section for more information on downloading the YOLO-NAS model for use in Frigate.

YOLO-NAS Setup & Config
warning

If you are using a Frigate+ YOLO-NAS model, you should not define any of the below model parameters in your config except for path. See the Frigate+ model docs for more information on setting up your model.

After placing the downloaded onnx model in your config folder, use the following configuration:

Navigate to Settings→System→Detector hardware and select ONNX from the detector type dropdown and click Add. Then navigate to Settings→System→Detection model and configure:

FieldValue
Object Detection Model Typeyolonas
Object detection model input width320 (should match whatever was set in notebook)
Object detection model input height320 (should match whatever was set in notebook)
Model Input Pixel Color Formatbgr
Model Input Tensor Shapenchw
Custom object detector model path/config/yolo_nas_s.onnx
Label map for custom object detector/labelmap/coco-80.txt

YOLO (v3, v4, v7, v9)​

YOLOv3, YOLOv4, YOLOv7, and YOLOv9 models are supported, but not included by default.

tip

The YOLO detector has been designed to support YOLOv3, YOLOv4, YOLOv7, and YOLOv9 models, but may support other YOLO model architectures as well. See the models section for more information on downloading YOLO models for use in Frigate.

YOLOv Setup & Config
warning

If you are using a Frigate+ model, you should not define any of the below model parameters in your config except for path. See the Frigate+ model docs for more information on setting up your model.

After placing the downloaded onnx model in your config folder, use the following configuration:

Navigate to Settings→System→Detector hardware and select ONNX from the detector type dropdown and click Add. Then navigate to Settings→System→Detection model and configure:

FieldValue
Object Detection Model Typeyolo-generic
Object detection model input width320 (should match the imgsize set during model export)
Object detection model input height320 (should match the imgsize set during model export)
Model Input Tensor Shapenchw
Model Input D Typefloat
Custom object detector model path/config/model_cache/yolo.onnx
Label map for custom object detector/labelmap/coco-80.txt

Note that the labelmap uses a subset of the complete COCO label set that has only 80 objects.

YOLOx​

YOLOx models are supported, but not included by default. See the models section for more information on downloading the YOLOx model for use in Frigate.

YOLOx Setup & Config

After placing the downloaded onnx model in your config folder, use the following configuration:

Navigate to Settings→System→Detector hardware and select ONNX from the detector type dropdown and click Add. Then navigate to Settings→System→Detection model and configure:

FieldValue
Object Detection Model Typeyolox
Object detection model input width416 (should match the imgsize set during model export)
Object detection model input height416 (should match the imgsize set during model export)
Model Input Tensor Shapenchw
Model Input D Typefloat_denorm
Custom object detector model path/config/model_cache/yolox_tiny.onnx
Label map for custom object detector/labelmap/coco-80.txt

Note that the labelmap uses a subset of the complete COCO label set that has only 80 objects.

RF-DETR​

RF-DETR is a DETR based model. The ONNX exported models are supported, but not included by default. See the models section for more information on downloading the RF-DETR model for use in Frigate.

RF-DETR Setup & Config

After placing the downloaded onnx model in your config/model_cache folder, use the following configuration:

Navigate to Settings→System→Detector hardware and select ONNX from the detector type dropdown and click Add. Then navigate to Settings→System→Detection model and configure:

FieldValue
Object Detection Model Typerfdetr
Object detection model input width320
Object detection model input height320
Model Input Tensor Shapenchw
Model Input D Typefloat
Custom object detector model path/config/model_cache/rfdetr.onnx

D-FINE / DEIMv2​

D-FINE and DEIMv2 are DETR based models that share the same ONNX input/output format. The ONNX exported models are supported, but not included by default. See the models section for downloading D-FINE or DEIMv2 for use in Frigate.

D-FINE Setup & Config

After placing the downloaded onnx model in your config/model_cache folder, use the following configuration:

Navigate to Settings→System→Detector hardware and select ONNX from the detector type dropdown and click Add. Then navigate to Settings→System→Detection model and configure:

FieldValue
Object Detection Model Typedfine
Object detection model input width640
Object detection model input height640
Model Input Tensor Shapenchw
Model Input D Typefloat
Custom object detector model path/config/model_cache/dfine_m_obj2coco.onnx
Label map for custom object detector/labelmap/coco-80.txt
DEIMv2 Setup & Config

After placing the downloaded onnx model in your config/model_cache folder, you can use the following configuration:

detectors:
  onnx:
    type: onnx

model:
  model_type: dfine
  width: 640
  height: 640
  input_tensor: nchw
  input_dtype: float
  path: /config/model_cache/deimv2_hgnetv2_n.onnx
  labelmap_path: /labelmap/coco-80.txt

Note that the labelmap uses a subset of the complete COCO label set that has only 80 objects.

The CPU detector type runs a TensorFlow Lite model utilizing the CPU without hardware acceleration. It is recommended to use a hardware accelerated detector type instead for better performance. To configure a CPU based detector, set the "type" attribute to "cpu".

danger

The CPU detector is not recommended for general use. If you do not have GPU or Edge TPU hardware, using the OpenVINO Detector in CPU mode is often more efficient than using the CPU detector.

The number of threads used by the interpreter can be specified using the "num_threads" attribute, and defaults to 3.

A TensorFlow Lite model is provided in the container at /cpu_model.tflite and is used by this detector type by default. To provide your own model, bind mount the file into the container and provide the path with model.path.

Navigate to Settings→System→Detector hardware and select CPU from the detector type dropdown and click Add. Configure the number of threads and click Add again to add additional CPU detectors as needed (one per camera is recommended).

When using CPU detectors, you can add one CPU detector per camera. Adding more detectors than the number of cameras should not improve performance.

Deepstack / CodeProject.AI Server Detector​

The Deepstack / CodeProject.AI Server detector for Frigate allows you to integrate Deepstack and CodeProject.AI object detection capabilities into Frigate. CodeProject.AI and DeepStack are open-source AI platforms that can be run on various devices such as the Raspberry Pi, Nvidia Jetson, and other compatible hardware. It is important to note that the integration is performed over the network, so the inference times may not be as fast as native Frigate detectors, but it still provides an efficient and reliable solution for object detection and tracking.

Setup​

To get started with CodeProject.AI, visit their official website to follow the instructions to download and install the AI server on your preferred device. Detailed setup instructions for CodeProject.AI are outside the scope of the Frigate documentation.

To integrate CodeProject.AI into Frigate, configure the detector as follows:

Navigate to Settings→System→Detector hardware and select DeepStack from the detector type dropdown and click Add. Set the API URL to point to your CodeProject.AI server (e.g., http://<your_codeproject_ai_server_ip>:<port>/v1/vision/detection).

Replace <your_codeproject_ai_server_ip> and <port> with the IP address and port of your CodeProject.AI server.

To verify that the integration is working correctly, start Frigate and observe the logs for any error messages related to CodeProject.AI. Additionally, you can check the Frigate web interface to see if the objects detected by CodeProject.AI are being displayed and tracked properly.

Community Supported Detectors

MemryX MX3​

This detector is available for use with the MemryX MX3 accelerator M.2 module. Frigate supports the MX3 on compatible hardware platforms, providing efficient and high-performance object detection.

See the installation docs for information on configuring the MemryX hardware.

To configure a MemryX detector, simply set the type attribute to memryx and follow the configuration guide below.

Configuration​

To configure the MemryX detector, use the following example configuration:

Single PCIe MemryX MX3​

Navigate to Settings→System→Detector hardware and select MemryX from the detector type dropdown and click Add, then set device to PCIe:0.

Multiple PCIe MemryX MX3 Modules​

Navigate to Settings→System→Detector hardware and select MemryX from the detector type dropdown and click Add to add multiple detectors, specifying PCIe:0, PCIe:1, PCIe:2, etc. as the device for each.

Supported Models​

MemryX .dfp models are automatically downloaded at runtime, if enabled, to the container at /memryx_models/model_folder/.

YOLO-NAS​

The YOLO-NAS model included in this detector is downloaded from the Models Section and compiled to DFP with mx_nc.

Note: The default model for the MemryX detector is YOLO-NAS 320x320.

The input size for YOLO-NAS can be set to either 320x320 (default) or 640x640.

  • The default size of 320x320 is optimized for lower CPU usage and faster inference times.
Configuration​

Below is the recommended configuration for using the YOLO-NAS (small) model with the MemryX detector:

Navigate to Settings→System→Detector hardware and select MemryX from the detector type dropdown and click Add, then set device to PCIe:0. Then navigate to Settings→System→Detection model and configure:

FieldValue
Object Detection Model Typeyolonas
Object detection model input width320 (can be set to 640 for higher resolution)
Object detection model input height320 (can be set to 640 for higher resolution)
Model Input Tensor Shapenchw
Model Input D Typefloat
Label map for custom object detector/labelmap/coco-80.txt

YOLOv9​

The YOLOv9s model included in this detector is downloaded from the original GitHub like in the Models Section and compiled to DFP with mx_nc.

Configuration​

Below is the recommended configuration for using the YOLOv9 (small) model with the MemryX detector:

Navigate to Settings→System→Detector hardware and select MemryX from the detector type dropdown and click Add, then set device to PCIe:0. Then navigate to Settings→System→Detection model and configure:

FieldValue
Object Detection Model Typeyolo-generic
Object detection model input width320 (can be set to 640 for higher resolution)
Object detection model input height320 (can be set to 640 for higher resolution)
Model Input Tensor Shapenchw
Model Input D Typefloat
Label map for custom object detector/labelmap/coco-80.txt

YOLOX​

The model is sourced from the OpenCV Model Zoo and precompiled to DFP.

Configuration​

Below is the recommended configuration for using the YOLOX (small) model with the MemryX detector:

Navigate to Settings→System→Detector hardware and select MemryX from the detector type dropdown and click Add, then set device to PCIe:0. Then navigate to Settings→System→Detection model and configure:

FieldValue
Object Detection Model Typeyolox
Object detection model input width640
Object detection model input height640
Model Input Tensor Shapenchw
Model Input D Typefloat_denorm
Label map for custom object detector/labelmap/coco-80.txt

SSDLite MobileNet v2​

The model is sourced from the OpenMMLab Model Zoo and has been converted to DFP.

Configuration​

Below is the recommended configuration for using the SSDLite MobileNet v2 model with the MemryX detector:

Navigate to Settings→System→Detector hardware and select MemryX from the detector type dropdown and click Add, then set device to PCIe:0. Then navigate to Settings→System→Detection model and configure:

FieldValue
Object Detection Model Typessd
Object detection model input width320
Object detection model input height320
Model Input Tensor Shapenchw
Model Input D Typefloat
Label map for custom object detector/labelmap/coco-80.txt

Using a Custom Model​

To use your own custom model, first compile it into a .dfp file, which is the format used by MemryX.

Compile the Model​

Custom models must be compiled using MemryX SDK 2.1.

Before compiling your model, install the MemryX Neural Compiler tools from the Install Tools page on the host.

Note: It is recommended to compile the model on the host machine, or on another separate machine, rather than inside the Frigate Docker container. Installing the compiler inside Docker may conflict with container packages. It is recommended to create a Python virtual environment and install the compiler there.

Once the SDK 2.1 environment is set up, follow the MemryX Compiler documentation to compile your model.

Example:

mx_nc -m yolonas.onnx -c 4 --autocrop -v --dfp_fname yolonas.dfp

For detailed instructions on compiling models, refer to the MemryX Compiler docs and Tutorials.

Package the Compiled Model​

  1. Package your compiled model into a .zip file.

  2. The .zip file must contain the compiled .dfp file.

  3. Depending on the model, the compiler may also generate a cropped post-processing network. If present, it will be named with the suffix _post.onnx.

  4. Bind-mount the .zip file into the container and specify its path using model.path in your config.

  5. Update labelmap_path to match your custom model's labels.

# The detector automatically selects the default model if nothing is provided in the config.
#
# Optionally, you can specify a local model path as a .zip file to override the default.
# If a local path is provided and the file exists, it will be used instead of downloading.
#
# Example:
# path: /config/yolonas.zip
#
# The .zip file must contain:
# ├── yolonas.dfp          (a file ending with .dfp)
# └── yolonas_post.onnx    (optional; only if the model includes a cropped post-processing network)

NVidia TensorRT Detector​

Nvidia Jetson devices may be used for object detection using the TensorRT libraries. Due to the size of the additional libraries, this detector is only provided in images with the -tensorrt-jp6 tag suffix, e.g. ghcr.io/blakeblackshear/frigate:stable-tensorrt-jp6. This detector is designed to work with Yolo models for object detection.

Generate Models​

The model used for TensorRT must be preprocessed on the same hardware platform that they will run on. This means that each user must run additional setup to generate a model file for the TensorRT library. A script is included that will build several common models.

The Frigate image will generate model files during startup if the specified model is not found. Processed models are stored in the /config/model_cache folder. Typically the /config path is mapped to a directory on the host already and the model_cache does not need to be mapped separately unless the user wants to store it in a different location on the host.

By default, no models will be generated, but this can be overridden by specifying the YOLO_MODELS environment variable in Docker. One or more models may be listed in a comma-separated format, and each one will be generated. Models will only be generated if the corresponding {model}.trt file is not present in the model_cache folder, so you can force a model to be regenerated by deleting it from your Frigate data folder.

If you have a Jetson device with DLAs (Xavier or Orin), you can generate a model that will run on the DLA by appending -dla to your model name, e.g. specify YOLO_MODELS=yolov7-320-dla. The model will run on DLA0 (Frigate does not currently support DLA1). DLA-incompatible layers will fall back to running on the GPU.

If your GPU does not support FP16 operations, you can pass the environment variable USE_FP16=False to disable it.

Specific models can be selected by passing an environment variable to the docker run command or in your docker-compose.yml file. Use the form -e YOLO_MODELS=yolov4-416,yolov4-tiny-416 to select one or more model names. The models available are shown below.

Available Models
yolov3-288
yolov3-416
yolov3-608
yolov3-spp-288
yolov3-spp-416
yolov3-spp-608
yolov3-tiny-288
yolov3-tiny-416
yolov4-288
yolov4-416
yolov4-608
yolov4-csp-256
yolov4-csp-512
yolov4-p5-448
yolov4-p5-896
yolov4-tiny-288
yolov4-tiny-416
yolov4x-mish-320
yolov4x-mish-640
yolov7-tiny-288
yolov7-tiny-416
yolov7-640
yolov7-416
yolov7-320
yolov7x-640
yolov7x-320

An example docker-compose.yml fragment that converts the yolov4-608 and yolov7x-640 models would look something like this:

frigate:
  environment:
    - YOLO_MODELS=yolov7-320,yolov7x-640
    - USE_FP16=false

Configuration Parameters​

The TensorRT detector can be selected by specifying tensorrt as the model type. The GPU will need to be passed through to the docker container using the same methods described in the Hardware Acceleration section. If you pass through multiple GPUs, you can select which GPU is used for a detector with the device configuration parameter. The device parameter is an integer value of the GPU index, as shown by nvidia-smi within the container.

The TensorRT detector uses .trt model files that are located in /config/model_cache/tensorrt by default. These model path and dimensions used will depend on which model you have generated.

Use the config below to work with generated TRT models:

Navigate to Settings→System→Detector hardware and select TensorRT from the detector type dropdown and click Add, then set the device to 0 (the default GPU index). Then navigate to Settings→System→Detection model and configure:

FieldValue
Custom object detector model path/config/model_cache/tensorrt/yolov7-320.trt
Label map for custom object detector/labelmap/coco-80.txt
Model Input Tensor Shapenchw
Model Input Pixel Color Formatrgb
Object detection model input width320 (MUST match the chosen model, e.g., yolov7-320 -> 320)
Object detection model input height320 (MUST match the chosen model, e.g., yolov7-320 -> 320)

Synaptics​

Hardware accelerated object detection is supported on the following SoCs:

  • SL1680

This implementation uses the Synaptics model conversion, version v3.1.0.

This implementation is based on sdk v1.5.0.

See the installation docs for information on configuring the SL-series NPU hardware.

Configuration​

When configuring the Synap detector, you have to specify the model: a local path.

SSD Mobilenet​

A synap model is provided in the container at /mobilenet.synap and is used by this detector type by default. The model comes from Synap-release Github.

Use the model configuration shown below when using the synaptics detector with the default synap model:

Navigate to Settings→System→Detector hardware and select Synaptics from the detector type dropdown and click Add. Then navigate to Settings→System→Detection model and configure:

FieldValue
Custom object detector model path/synaptics/mobilenet.synap
Object detection model input width224
Object detection model input height224
Model Input Tensor Shapenhwc
Label map for custom object detector/labelmap/coco-80.txt

Rockchip platform​

Hardware accelerated object detection is supported on the following SoCs:

  • RK3562
  • RK3566
  • RK3568
  • RK3576
  • RK3588

This implementation uses the Rockchip's RKNN-Toolkit2, version v2.3.2.

info

If no custom model is provided, the RKNN detector downloads a default model from GitHub on first startup. Once cached, the model works fully offline. See Network Requirements for details.

tip

When using many cameras one detector may not be enough to keep up. Multiple detectors can be defined assuming NPU resources are available. An example configuration would be:

Navigate to Settings→System→Detector hardware and select RKNN from the detector type dropdown and click Add to add multiple detectors, each with num_cores set to 0 for automatic selection.

Prerequisites​

Make sure to follow the Rockchip specific installation instructions.

tip

You can get the load of your NPU with the following command:

$ cat /sys/kernel/debug/rknpu/load
>> NPU load:  Core0:  0%, Core1:  0%, Core2:  0%,

RockChip Supported Models​

This config.yml shows all relevant options to configure the detector and explains them. All values shown are the default values (except for two). Lines that are required at least to use the detector are labeled as required, all other lines are optional.

Navigate to Settings→System→Detector hardware and select RKNN from the detector type dropdown and click Add. Set num_cores to 0 for automatic selection (increase for better performance on multicore NPUs, e.g., set to 3 on rk3588).

The inference time was determined on a rk3588 with 3 NPU cores.

ModelSize in mbInference time in ms
deci-fp16-yolonas_s2425
deci-fp16-yolonas_m6235
deci-fp16-yolonas_l8145
frigate-fp16-yolov9-t635
rock-i8-yolox_nano314
rock-i8_yolox_tiny618
  • All models are automatically downloaded and stored in the folder config/model_cache/rknn_cache. After upgrading Frigate, you should remove older models to free up space.
  • You can also provide your own .rknn model. You should not save your own models in the rknn_cache folder, store them directly in the model_cache folder or another subfolder. To convert a model to .rknn format see the rknn-toolkit2 (requires a x86 machine). Note, that there is only post-processing for the supported models.

YOLO-NAS​

Navigate to Settings→System→Detection model and configure:

FieldValue
Custom object detector model pathdeci-fp16-yolonas_s (or deci-fp16-yolonas_m, deci-fp16-yolonas_l)
Object Detection Model Typeyolonas
Object detection model input width320
Object detection model input height320
Model Input Pixel Color Formatbgr
Model Input Tensor Shapenhwc
Label map for custom object detector/labelmap/coco-80.txt
warning

The pre-trained YOLO-NAS weights from DeciAI are subject to their license and can't be used commercially. For more information, see: https://docs.deci.ai/super-gradients/latest/LICENSE.YOLONAS.html

YOLO (v9)​

Navigate to Settings→System→Detection model and configure:

FieldValue
Custom object detector model pathfrigate-fp16-yolov9-t (or other yolov9 variants)
Object Detection Model Typeyolo-generic
Object detection model input width320
Object detection model input height320
Model Input Tensor Shapenhwc
Label map for custom object detector/labelmap/coco-80.txt

YOLOx​

Navigate to Settings→System→Detection model and configure:

FieldValue
Custom object detector model pathrock-i8-yolox_nano (or other yolox variants)
Object Detection Model Typeyolox
Object detection model input width416
Object detection model input height416
Model Input Tensor Shapenhwc
Label map for custom object detector/labelmap/coco-80.txt

Converting your own onnx model to rknn format​

To convert a onnx model to the rknn format using the rknn-toolkit2 you have to:

  • Place one ore more models in onnx format in the directory config/model_cache/rknn_cache/onnx on your docker host (this might require sudo privileges).
  • Save the configuration file under config/conv2rknn.yaml (see below for details).
  • Run docker exec <frigate_container_id> python3 /opt/conv2rknn.py. If the conversion was successful, the rknn models will be placed in config/model_cache/rknn_cache.

This is an example configuration file that you need to adjust to your specific onnx model:

soc: ["rk3562", "rk3566", "rk3568", "rk3576", "rk3588"]
quantization: false

output_name: "{input_basename}"

config:
  mean_values: [[0, 0, 0]]
  std_values: [[255, 255, 255]]
  quant_img_RGB2BGR: true

Explanation of the paramters:

  • soc: A list of all SoCs you want to build the rknn model for. If you don't specify this parameter, the script tries to find out your SoC and builds the rknn model for this one.
  • quantization: true: 8 bit integer (i8) quantization, false: 16 bit float (fp16). Default: false.
  • output_name: The output name of the model. The following variables are available:
    • quant: "i8" or "fp16" depending on the config
    • input_basename: the basename of the input model (e.g. "my_model" if the input model is calles "my_model.onnx")
    • soc: the SoC this model was build for (e.g. "rk3588")
    • tk_version: Version of rknn-toolkit2 (e.g. "2.3.0")
    • example: Specifying output_name = "frigate-{quant}-{input_basename}-{soc}-v{tk_version}" could result in a model called frigate-i8-my_model-rk3588-v2.3.0.rknn.
  • config: Configuration passed to rknn-toolkit2 for model conversion. For an explanation of all available parameters have a look at section "2.2. Model configuration" of this manual.

DeGirum​

DeGirum is a detector that can use any type of hardware listed on their website. DeGirum can be used with local hardware through a DeGirum AI Server, or through the use of @local. You can also connect directly to DeGirum's AI Hub to run inferences. Please Note: This detector cannot be used for commercial purposes.

Configuration​

AI Server Inference​

Before starting with the config file for this section, you must first launch an AI server. DeGirum has an AI server ready to use as a docker container. Add this to your docker-compose.yml to get started:

degirum_detector:
  container_name: degirum
  image: degirum/aiserver:latest
  privileged: true
  ports:
    - "8778:8778"

All supported hardware will automatically be found on your AI server host as long as relevant runtimes and drivers are properly installed on your machine. Refer to DeGirum's docs site if you have any trouble.

Once completed, configure the detector as follows:

Navigate to Settings→System→Detector hardware and select DeGirum from the detector type dropdown and click Add. Set the location to your AI server (e.g., service name, container name, or host:port), the zoo to degirum/public, and provide your authentication token if needed.

Setting up a model in the config.yml is similar to setting up an AI server. You can set it to:

  • A model listed on the AI Hub, given that the correct zoo name is listed in your detector
    • If this is what you choose to do, the correct model will be downloaded onto your machine before running.
  • A local directory acting as a zoo. See DeGirum's docs site for more information.
  • A path to some model.json.
model:
  path: ./mobilenet_v2_ssd_coco--300x300_quant_n2x_orca1_1 # directory to model .json and file
  width: 300 # width is in the model name as the first number in the "int"x"int" section
  height: 300 # height is in the model name as the second number in the "int"x"int" section
  input_pixel_format: rgb/bgr # look at the model.json to figure out which to put here

Local Inference​

It is also possible to eliminate the need for an AI server and run the hardware directly. The benefit of this approach is that you eliminate any bottlenecks that occur when transferring prediction results from the AI server docker container to the frigate one. However, the method of implementing local inference is different for every device and hardware combination, so it's usually more trouble than it's worth. A general guideline to achieve this would be:

  1. Ensuring that the frigate docker container has the runtime you want to use. So for instance, running @local for Hailo means making sure the container you're using has the Hailo runtime installed.
  2. To double check the runtime is detected by the DeGirum detector, make sure the degirum sys-info command properly shows whatever runtimes you mean to install.
  3. Create a DeGirum detector in your configuration.

Navigate to Settings→System→Detector hardware and select DeGirum from the detector type dropdown and click Add. Set the location to @local, the zoo to degirum/public, and provide your authentication token.

Once degirum_detector is setup, you can choose a model through 'model' section in the config.yml file.

model:
  path: mobilenet_v2_ssd_coco--300x300_quant_n2x_orca1_1
  width: 300 # width is in the model name as the first number in the "int"x"int" section
  height: 300 # height is in the model name as the second number in the "int"x"int" section
  input_pixel_format: rgb/bgr # look at the model.json to figure out which to put here

AI Hub Cloud Inference​

If you do not possess whatever hardware you want to run, there's also the option to run cloud inferences. Do note that your detection fps might need to be lowered as network latency does significantly slow down this method of detection. For use with Frigate, we highly recommend using a local AI server as described above. To set up cloud inferences,

  1. Sign up at DeGirum's AI Hub.
  2. Get an access token.
  3. Create a DeGirum detector in your configuration.

Navigate to Settings→System→Detector hardware and select DeGirum from the detector type dropdown and click Add. Set the location to @cloud, the zoo to degirum/public, and provide your authentication token.

Once degirum_detector is setup, you can choose a model through 'model' section in the config.yml file.

model:
  path: mobilenet_v2_ssd_coco--300x300_quant_n2x_orca1_1
  width: 300 # width is in the model name as the first number in the "int"x"int" section
  height: 300 # height is in the model name as the second number in the "int"x"int" section
  input_pixel_format: rgb/bgr # look at the model.json to figure out which to put here

AXERA​

Hardware accelerated object detection is supported on the following SoCs:

  • AX650N
  • AX8850N

This implementation uses the AXera Pulsar2 Toolchain.

See the installation docs for information on configuring the AXEngine hardware.

info

The AXEngine detector downloads its default model from HuggingFace on first startup. Once cached, the model works fully offline. See Network Requirements for details.

Configuration​

When configuring the AXEngine detector, you have to specify the model name.

yolov9​

A yolov9 model is provided in the container at /axmodels and is used by this detector type by default.

Use the model configuration shown below when using the axengine detector with the default axmodel:

Navigate to Settings→System→Detector hardware and select AXEngine NPU from the detector type dropdown and click Add. Then navigate to Settings→System→Detection model and configure:

FieldValue
Custom object detector model pathfrigate-yolov9-tiny
Object Detection Model Typeyolo-generic
Object detection model input width320
Object detection model input height320
Model Input D Typeint
Model Input Pixel Color Formatbgr
Label map for custom object detector/labelmap/coco-80.txt

Models

Some model types are not included in Frigate by default.

Downloading Models​

Here are some tips for getting different model types

Downloading D-FINE Model​

D-FINE can be exported as ONNX by running the command below. You can copy and paste the whole thing to your terminal and execute, altering MODEL_SIZE=s in the first line to s, m, or l size.

docker build . --build-arg MODEL_SIZE=s --output . -f- <<'EOF'
FROM python:3.11 AS build
RUN apt-get update && apt-get install --no-install-recommends -y libgl1 && rm -rf /var/lib/apt/lists/*
COPY --from=ghcr.io/astral-sh/uv:0.8.0 /uv /bin/
WORKDIR /dfine
RUN git clone https://github.com/Peterande/D-FINE.git .
RUN uv pip install --system -r requirements.txt
RUN uv pip install --system onnx onnxruntime onnxsim onnxscript
# Create output directory and download checkpoint
RUN mkdir -p output
ARG MODEL_SIZE
RUN wget https://github.com/Peterande/storage/releases/download/dfinev1.0/dfine_${MODEL_SIZE}_obj2coco.pth -O output/dfine_${MODEL_SIZE}_obj2coco.pth
# Modify line 58 of export_onnx.py to change batch size to 1
RUN sed -i '58s/data = torch.rand(.*)/data = torch.rand(1, 3, 640, 640)/' tools/deployment/export_onnx.py
RUN python3 tools/deployment/export_onnx.py -c configs/dfine/objects365/dfine_hgnetv2_${MODEL_SIZE}_obj2coco.yml -r output/dfine_${MODEL_SIZE}_obj2coco.pth
FROM scratch
ARG MODEL_SIZE
COPY --from=build /dfine/output/dfine_${MODEL_SIZE}_obj2coco.onnx /dfine-${MODEL_SIZE}.onnx
EOF

Downloading DEIMv2 Model​

DEIMv2 can be exported as ONNX by running the command below. Pretrained weights are available on Hugging Face for two backbone families:

  • HGNetv2 (smaller/faster): atto, femto, pico, n
  • DINOv3 (larger/more accurate): s, m, l, x

Set BACKBONE and MODEL_SIZE in the first line to match your desired variant. Hugging Face model names use uppercase (e.g. HGNetv2_N, DINOv3_S), while config files use lowercase (e.g. hgnetv2_n, dinov3_s).

docker build . --rm --build-arg BACKBONE=hgnetv2 --build-arg MODEL_SIZE=n --output . -f- <<'EOF'
FROM python:3.11-slim AS build
RUN apt-get update && apt-get install --no-install-recommends -y git libgl1 libglib2.0-0 && rm -rf /var/lib/apt/lists/*
COPY --from=ghcr.io/astral-sh/uv:0.8.0 /uv /bin/
WORKDIR /deimv2
RUN git clone https://github.com/Intellindust-AI-Lab/DEIMv2.git .
# Install CPU-only PyTorch first to avoid pulling CUDA variant
RUN uv pip install --no-cache --system torch torchvision --index-url https://download.pytorch.org/whl/cpu
RUN uv pip install --no-cache --system -r requirements.txt
RUN uv pip install --no-cache --system onnx safetensors huggingface_hub
RUN mkdir -p output
ARG BACKBONE
ARG MODEL_SIZE
# Download from Hugging Face and convert safetensors to pth
RUN python3 -c "\
from huggingface_hub import hf_hub_download; \
from safetensors.torch import load_file; \
import torch; \
backbone = '${BACKBONE}'.replace('hgnetv2','HGNetv2').replace('dinov3','DINOv3'); \
size = '${MODEL_SIZE}'.upper(); \
st = load_file(hf_hub_download('Intellindust/DEIMv2_' + backbone + '_' + size + '_COCO', 'model.safetensors')); \
torch.save({'model': st}, 'output/deimv2.pth')"
RUN sed -i "s/data = torch.rand(2/data = torch.rand(1/" tools/deployment/export_onnx.py
# HuggingFace safetensors omits frozen constants that the model constructor initializes
RUN sed -i "s/cfg.model.load_state_dict(state)/cfg.model.load_state_dict(state, strict=False)/" tools/deployment/export_onnx.py
RUN python3 tools/deployment/export_onnx.py -c configs/deimv2/deimv2_${BACKBONE}_${MODEL_SIZE}_coco.yml -r output/deimv2.pth
FROM scratch
ARG BACKBONE
ARG MODEL_SIZE
COPY --from=build /deimv2/output/deimv2.onnx /deimv2_${BACKBONE}_${MODEL_SIZE}.onnx
EOF

Downloading RF-DETR Model​

RF-DETR can be exported as ONNX by running the command below. You can copy and paste the whole thing to your terminal and execute, altering MODEL_SIZE=Nano in the first line to Nano, Small, or Medium size.

docker build . --build-arg MODEL_SIZE=Nano --rm --output . -f- <<'EOF'
FROM python:3.12 AS build
RUN apt-get update && apt-get install --no-install-recommends -y libgl1 && rm -rf /var/lib/apt/lists/*
COPY --from=ghcr.io/astral-sh/uv:0.10.4 /uv /bin/
WORKDIR /rfdetr
RUN uv pip install --system rfdetr[onnxexport] torch==2.8.0 onnx==1.19.1 transformers==4.57.6 onnxscript
ARG MODEL_SIZE
RUN python3 -c "from rfdetr import RFDETR${MODEL_SIZE}; x = RFDETR${MODEL_SIZE}(resolution=320); x.export(simplify=True)"
FROM scratch
ARG MODEL_SIZE
COPY --from=build /rfdetr/output/inference_model.onnx /rfdetr-${MODEL_SIZE}.onnx
EOF

Downloading YOLO-NAS Model​

You can build and download a compatible model with pre-trained weights using this notebook Open In Colab which can be run directly in Google Colab.

warning

The pre-trained YOLO-NAS weights from DeciAI are subject to their license and can't be used commercially. For more information, see: https://docs.deci.ai/super-gradients/latest/LICENSE.YOLONAS.html

The input image size in this notebook is set to 320x320. This results in lower CPU usage and faster inference times without impacting performance in most cases due to the way Frigate crops video frames to areas of interest before running detection. The notebook and config can be updated to 640x640 if desired.

Downloading YOLO Models​

YOLOx​

YOLOx models can be downloaded from the YOLOx repo.

YOLOv3, YOLOv4, and YOLOv7​

To export as ONNX:

git clone https://github.com/NateMeyer/tensorrt_demos
cd tensorrt_demos/yolo
./download_yolo.sh
python3 yolo_to_onnx.py -m yolov7-320

YOLOv9 for Google Coral Support​

Download the model, bind mount the file into the container, and provide the path with model.path. Note that the linked model requires a 17-label labelmap file that includes only 17 COCO classes.

YOLOv9 for other detectors​

YOLOv9 model can be exported as ONNX using the command below. You can copy and paste the whole thing to your terminal and execute, altering MODEL_SIZE=t and IMG_SIZE=320 in the first line to the model size you would like to convert (available model sizes are t, s, m, c, and e, common image sizes are 320 and 640).

docker build . --build-arg MODEL_SIZE=t --build-arg IMG_SIZE=320 --output . -f- <<'EOF'
FROM python:3.11 AS build
RUN apt-get update && apt-get install --no-install-recommends -y cmake libgl1 && rm -rf /var/lib/apt/lists/*
COPY --from=ghcr.io/astral-sh/uv:0.10.4 /uv /bin/
WORKDIR /yolov9
ADD https://github.com/WongKinYiu/yolov9.git .
RUN uv pip install --system -r requirements.txt
RUN uv pip install --system onnx==1.18.0 onnxruntime onnx-simplifier==0.4.* onnxscript
ARG MODEL_SIZE
ARG IMG_SIZE
ADD https://github.com/WongKinYiu/yolov9/releases/download/v0.1/yolov9-${MODEL_SIZE}-converted.pt yolov9-${MODEL_SIZE}.pt
RUN sed -i "s/ckpt = torch.load(attempt_download(w), map_location='cpu')/ckpt = torch.load(attempt_download(w), map_location='cpu', weights_only=False)/g" models/experimental.py
RUN python3 export.py --weights ./yolov9-${MODEL_SIZE}.pt --imgsz ${IMG_SIZE} --simplify --include onnx
FROM scratch
ARG MODEL_SIZE
ARG IMG_SIZE
COPY --from=build /yolov9/yolov9-${MODEL_SIZE}.onnx /yolov9-${MODEL_SIZE}-${IMG_SIZE}.onnx
EOF