Nvidia Jetson ffmpeg + TensorRT support (#6458)

* Non-Jetson changes

Required for later commits:
- Allow base image to be overridden (and don't assume its WORKDIR)
- Ensure python3.9
- Map hwaccel decode presets as strings instead of lists
Not required:
- Fix existing documentation
- Simplify hwaccel scale logic

* Prepare for multi-arch tensorrt build

* Add tensorrt images for Jetson boards

* Add Jetson ffmpeg hwaccel

* Update docs

* Add CODEOWNERS

* CI

* Change default model from yolov7-tiny-416 to yolov7-320

In my experience the tiny models perform markedly worse without being
much faster

* fixup! Update docs

.github/workflows/ci.yml

@@ -79,7 +79,7 @@ jobs:
           set: |
             rpi.tags=ghcr.io/${{ steps.lowercaseRepo.outputs.lowercase }}:${{ github.ref_name }}-${{ env.SHORT_SHA }}-rpi
             *.cache-from=type=gha
-      - name: Build and push TensorRT
+      - name: Build and push TensorRT (x86 GPU)
         uses: docker/bake-action@v3
         with:
           push: true
@@ -88,6 +88,34 @@ jobs:
           set: |
             tensorrt.tags=ghcr.io/${{ steps.lowercaseRepo.outputs.lowercase }}:${{ github.ref_name }}-${{ env.SHORT_SHA }}-tensorrt
             *.cache-from=type=gha
+      - name: Build and push TensorRT (Jetson, Jetpack 4)
+        env:
+          ARCH: arm64
+          BASE_IMAGE: timongentzsch/l4t-ubuntu20-opencv:latest
+          SLIM_BASE: timongentzsch/l4t-ubuntu20-opencv:latest
+          TRT_BASE: timongentzsch/l4t-ubuntu20-opencv:latest
+        uses: docker/bake-action@v3
+        with:
+          push: true
+          targets: tensorrt
+          files: docker/tensorrt/trt.hcl
+          set: |
+            tensorrt.tags=ghcr.io/${{ steps.lowercaseRepo.outputs.lowercase }}:${{ github.ref_name }}-${{ env.SHORT_SHA }}-tensorrt-jp4
+            *.cache-from=type=gha
+      - name: Build and push TensorRT (Jetson, Jetpack 5)
+        env:
+          ARCH: arm64
+          BASE_IMAGE: nvcr.io/nvidia/l4t-tensorrt:r8.5.2-runtime
+          SLIM_BASE: nvcr.io/nvidia/l4t-tensorrt:r8.5.2-runtime
+          TRT_BASE: nvcr.io/nvidia/l4t-tensorrt:r8.5.2-runtime
+        uses: docker/bake-action@v3
+        with:
+          push: true
+          targets: tensorrt
+          files: docker/tensorrt/trt.hcl
+          set: |
+            tensorrt.tags=ghcr.io/${{ steps.lowercaseRepo.outputs.lowercase }}:${{ github.ref_name }}-${{ env.SHORT_SHA }}-tensorrt-jp5
+            *.cache-from=type=gha
       - name: Assemble and push default build
         uses: int128/docker-manifest-create-action@v1
         with:

CODEOWNERS

@@ -0,0 +1,2 @@
+# Community-supported boards
+/docker/tensorrt/ @madsciencetist @NateMeyer

docker-compose.yml

@@ -21,6 +21,8 @@ services:
                       - driver: nvidia
                         count: 1
                         capabilities: [gpu]
+    environment:
+      YOLO_MODELS: yolov7-320
     devices:
       - /dev/bus/usb:/dev/bus/usb
       # - /dev/dri:/dev/dri # for intel hwaccel, needs to be updated for your hardware
@@ -30,8 +32,6 @@ services:
       - /etc/localtime:/etc/localtime:ro
       - ./config:/config
       - ./debug:/media/frigate
-      # Create the trt-models folder using the documented method of generating TRT models
-      # - ./debug/trt-models:/trt-models
       - /dev/bus/usb:/dev/bus/usb
   mqtt:
     container_name: mqtt

docker/main/Dockerfile

@@ -3,11 +3,14 @@
 # https://askubuntu.com/questions/972516/debian-frontend-environment-variable
 ARG DEBIAN_FRONTEND=noninteractive
 
-FROM debian:11 AS base
+ARG BASE_IMAGE=debian:11
+ARG SLIM_BASE=debian:11-slim
+
+FROM ${BASE_IMAGE} AS base
 
 FROM --platform=linux/amd64 debian:11 AS base_amd64
 
-FROM debian:11-slim AS slim-base
+FROM ${SLIM_BASE} AS slim-base
 
 FROM slim-base AS wget
 ARG DEBIAN_FRONTEND
@@ -123,8 +126,8 @@ RUN apt-get -qq update \
     && echo "deb http://deb.debian.org/debian bullseye main contrib non-free" | tee /etc/apt/sources.list.d/raspi.list \
     && apt-get -qq update \
     && apt-get -qq install -y \
-    python3 \
+    python3.9 \
-    python3-dev \
+    python3.9-dev \
     wget \
     # opencv dependencies
     build-essential cmake git pkg-config libgtk-3-dev \
@@ -137,14 +140,17 @@ RUN apt-get -qq update \
     gcc gfortran libopenblas-dev liblapack-dev && \
     rm -rf /var/lib/apt/lists/*
 
+# Ensure python3 defaults to python3.9
+RUN update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.9 1
+
 RUN wget -q https://bootstrap.pypa.io/get-pip.py -O get-pip.py \
     && python3 get-pip.py "pip"
 
 COPY docker/main/requirements.txt /requirements.txt
-RUN pip3 install -r requirements.txt
+RUN pip3 install -r /requirements.txt
 
 COPY docker/main/requirements-wheels.txt /requirements-wheels.txt
-RUN pip3 wheel --wheel-dir=/wheels -r requirements-wheels.txt
+RUN pip3 wheel --wheel-dir=/wheels -r /requirements-wheels.txt
 
 
 # Collect deps in a single layer
@@ -176,6 +182,7 @@ RUN --mount=type=bind,source=docker/main/install_deps.sh,target=/deps/install_de
     /deps/install_deps.sh
 
 RUN --mount=type=bind,from=wheels,source=/wheels,target=/deps/wheels \
+    python3 -m pip install --upgrade pip && \
     pip3 install -U /deps/wheels/*.whl
 
 COPY --from=deps-rootfs / /

docker/main/install_deps.sh

@@ -10,11 +10,15 @@ apt-get -qq install --no-install-recommends -y \
     wget \
     procps vainfo \
     unzip locales tzdata libxml2 xz-utils \
+    python3.9 \
     python3-pip \
     curl \
     jq \
     nethogs
 
+# ensure python3 defaults to python3.9
+update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.9 1
+
 mkdir -p -m 600 /root/.gnupg
 
 # add coral repo
@@ -23,8 +27,10 @@ curl -fsSLo - https://packages.cloud.google.com/apt/doc/apt-key.gpg | \
 echo "deb https://packages.cloud.google.com/apt coral-edgetpu-stable main" | tee /etc/apt/sources.list.d/coral-edgetpu.list
 echo "libedgetpu1-max libedgetpu/accepted-eula select true" | debconf-set-selections
 
-# enable non-free repo
+# enable non-free repo in Debian
-sed -i -e's/ main/ main contrib non-free/g' /etc/apt/sources.list
+if grep -q "Debian" /etc/issue; then
+    sed -i -e's/ main/ main contrib non-free/g' /etc/apt/sources.list
+fi
 
 # coral drivers
 apt-get -qq update

docker/tensorrt/Dockerfile.amd64

@@ -9,28 +9,11 @@ ARG DEBIAN_FRONTEND
 ARG TARGETARCH
 
 # Add TensorRT wheels to another folder
-COPY docker/tensorrt/requirements.txt /requirements-tensorrt.txt
+COPY docker/tensorrt/requirements-amd64.txt /requirements-tensorrt.txt
 RUN mkdir -p /trt-wheels && pip3 wheel --wheel-dir=/trt-wheels -r /requirements-tensorrt.txt
 
-# Build TensorRT-specific library
+FROM tensorrt-base AS frigate-tensorrt
-FROM nvcr.io/nvidia/tensorrt:23.03-py3 AS trt-deps
-
-RUN --mount=type=bind,source=docker/tensorrt/detector/tensorrt_libyolo.sh,target=/tensorrt_libyolo.sh \
-    /tensorrt_libyolo.sh
-
-# Frigate w/ TensorRT Support as separate image
-FROM deps AS frigate-tensorrt
-
-#Disable S6 Global timeout
-ENV S6_CMD_WAIT_FOR_SERVICES_MAXTIME=0
-
 ENV TRT_VER=8.5.3
-ENV YOLO_MODELS="yolov7-tiny-416"
-
-COPY --from=trt-deps /usr/local/lib/libyolo_layer.so /usr/local/lib/libyolo_layer.so
-COPY --from=trt-deps /usr/local/src/tensorrt_demos /usr/local/src/tensorrt_demos
-COPY docker/tensorrt/detector/rootfs/ /
-
 RUN --mount=type=bind,from=trt-wheels,source=/trt-wheels,target=/deps/trt-wheels \
     pip3 install -U /deps/trt-wheels/*.whl && \
     ldconfig

docker/tensorrt/Dockerfile.arm64

@@ -0,0 +1,79 @@
+# syntax=docker/dockerfile:1.4
+
+# https://askubuntu.com/questions/972516/debian-frontend-environment-variable
+ARG DEBIAN_FRONTEND=noninteractive
+
+ARG BASE_IMAGE
+FROM ${BASE_IMAGE} AS build-wheels
+ARG DEBIAN_FRONTEND
+
+# Use a separate container to build wheels to prevent build dependencies in final image
+RUN apt-get -qq update \
+    && apt-get -qq install -y --no-install-recommends \
+       python3.9 python3.9-dev \
+       wget build-essential cmake git \
+    && rm -rf /var/lib/apt/lists/*
+
+# Ensure python3 defaults to python3.9
+RUN update-alternatives --install /usr/bin/python3 python3 /usr/bin/python3.9 1
+
+RUN wget -q https://bootstrap.pypa.io/get-pip.py -O get-pip.py \
+    && python3 get-pip.py "pip"
+
+
+FROM build-wheels AS trt-wheels
+ARG DEBIAN_FRONTEND
+ARG TARGETARCH
+
+# python-tensorrt build deps are 3.4 GB!
+RUN apt-get update \
+    && apt-get install -y ccache cuda-cudart-dev-* cuda-nvcc-* libnvonnxparsers-dev libnvparsers-dev libnvinfer-plugin-dev \
+    && ([ -e /usr/local/cuda ] || ln -s /usr/local/cuda-* /usr/local/cuda) \
+    && rm -rf /var/lib/apt/lists/*;
+
+# Determine version of tensorrt already installed in base image, e.g. "Version: 8.4.1-1+cuda11.4"
+RUN NVINFER_VER=$(dpkg -s libnvinfer8 | grep -Po "Version: \K.*") \
+    && echo $NVINFER_VER | grep -Po "^\d+\.\d+\.\d+" > /etc/TENSORRT_VER
+
+RUN --mount=type=bind,source=docker/tensorrt/detector/build_python_tensorrt.sh,target=/deps/build_python_tensorrt.sh \
+    --mount=type=cache,target=/root/.ccache \
+    export PATH="/usr/lib/ccache:$PATH" CCACHE_DIR=/root/.ccache CCACHE_MAXSIZE=2G \
+    && TENSORRT_VER=$(cat /etc/TENSORRT_VER) /deps/build_python_tensorrt.sh
+
+COPY docker/tensorrt/requirements-arm64.txt /requirements-tensorrt.txt
+RUN pip3 wheel --wheel-dir=/trt-wheels -r /requirements-tensorrt.txt
+
+FROM build-wheels AS trt-model-wheels
+ARG DEBIAN_FRONTEND
+
+RUN apt-get update \
+    && apt-get install -y protobuf-compiler libprotobuf-dev \
+    && rm -rf /var/lib/apt/lists/*
+RUN --mount=type=bind,source=docker/tensorrt/requirements-models-arm64.txt,target=/requirements-tensorrt-models.txt \
+    pip3 wheel --wheel-dir=/trt-model-wheels -r /requirements-tensorrt-models.txt
+
+FROM wget AS jetson-ffmpeg
+ARG DEBIAN_FRONTEND
+ENV CCACHE_DIR /root/.ccache
+ENV CCACHE_MAXSIZE 2G
+RUN --mount=type=bind,source=docker/tensorrt/build_jetson_ffmpeg.sh,target=/deps/build_jetson_ffmpeg.sh \
+    --mount=type=cache,target=/root/.ccache \
+    /deps/build_jetson_ffmpeg.sh
+
+# Frigate w/ TensorRT for NVIDIA Jetson platforms
+FROM tensorrt-base AS frigate-tensorrt
+RUN apt-get update \
+    && apt-get install -y python-is-python3 libprotobuf17  \
+    && rm -rf /var/lib/apt/lists/*
+
+RUN rm -rf /usr/lib/btbn-ffmpeg/
+COPY --from=jetson-ffmpeg /rootfs /
+
+COPY --from=trt-wheels /etc/TENSORRT_VER /etc/TENSORRT_VER
+RUN --mount=type=bind,from=trt-wheels,source=/trt-wheels,target=/deps/trt-wheels \
+    --mount=type=bind,from=trt-model-wheels,source=/trt-model-wheels,target=/deps/trt-model-wheels \
+    pip3 install -U /deps/trt-wheels/*.whl /deps/trt-model-wheels/*.whl \
+    && ldconfig
+
+WORKDIR /opt/frigate/
+COPY --from=rootfs / /

docker/tensorrt/Dockerfile.base

@@ -0,0 +1,26 @@
+# syntax=docker/dockerfile:1.4
+
+# https://askubuntu.com/questions/972516/debian-frontend-environment-variable
+ARG DEBIAN_FRONTEND=noninteractive
+
+ARG TRT_BASE=nvcr.io/nvidia/tensorrt:23.03-py3
+
+# Build TensorRT-specific library
+FROM ${TRT_BASE} AS trt-deps
+
+RUN apt-get update \
+    && apt-get install -y git build-essential cuda-nvcc-* cuda-nvtx-* libnvinfer-dev libnvinfer-plugin-dev libnvparsers-dev libnvonnxparsers-dev \
+    && rm -rf /var/lib/apt/lists/*
+RUN --mount=type=bind,source=docker/tensorrt/detector/tensorrt_libyolo.sh,target=/tensorrt_libyolo.sh \
+    /tensorrt_libyolo.sh
+
+# Frigate w/ TensorRT Support as separate image
+FROM deps AS tensorrt-base
+
+#Disable S6 Global timeout
+ENV S6_CMD_WAIT_FOR_SERVICES_MAXTIME=0
+
+COPY --from=trt-deps /usr/local/lib/libyolo_layer.so /usr/local/lib/libyolo_layer.so
+COPY --from=trt-deps /usr/local/src/tensorrt_demos /usr/local/src/tensorrt_demos
+COPY docker/tensorrt/detector/rootfs/ /
+ENV YOLO_MODELS="yolov7-320"

docker/tensorrt/build_jetson_ffmpeg.sh

@@ -0,0 +1,59 @@
+#!/bin/bash
+
+# For jetson platforms, build ffmpeg with custom patches. NVIDIA supplies a deb
+# with accelerated decoding, but it doesn't have accelerated scaling or encoding
+
+set -euxo pipefail
+
+INSTALL_PREFIX=/rootfs/usr/local
+
+apt-get -qq update
+apt-get -qq install -y --no-install-recommends build-essential ccache clang cmake pkg-config
+apt-get -qq install -y --no-install-recommends libx264-dev libx265-dev
+
+pushd /tmp
+
+# Install libnvmpi to enable nvmpi decoders (h264_nvmpi, hevc_nvmpi)
+if [ -e /usr/local/cuda-10.2 ]; then
+    # assume Jetpack 4.X
+    wget -q https://developer.nvidia.com/embedded/L4T/r32_Release_v5.0/T186/Jetson_Multimedia_API_R32.5.0_aarch64.tbz2 -O jetson_multimedia_api.tbz2
+else
+    # assume Jetpack 5.X
+    wget -q https://developer.nvidia.com/downloads/embedded/l4t/r35_release_v3.1/release/jetson_multimedia_api_r35.3.1_aarch64.tbz2 -O jetson_multimedia_api.tbz2
+fi
+tar xaf jetson_multimedia_api.tbz2 -C / && rm jetson_multimedia_api.tbz2
+
+wget -q https://github.com/madsciencetist/jetson-ffmpeg/archive/refs/heads/master.zip
+unzip master.zip && rm master.zip && cd jetson-ffmpeg-master
+LD_LIBRARY_PATH=$(pwd)/stubs:$LD_LIBRARY_PATH   # tegra multimedia libs aren't available in image, so use stubs for ffmpeg build
+mkdir build
+cd build
+cmake .. -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=$INSTALL_PREFIX
+make -j$(nproc)
+make install
+cd ../../
+
+# Install nv-codec-headers to enable ffnvcodec filters (scale_cuda)
+wget -q https://github.com/FFmpeg/nv-codec-headers/archive/refs/heads/master.zip
+unzip master.zip && rm master.zip && cd nv-codec-headers-master
+make PREFIX=$INSTALL_PREFIX install
+cd ../ && rm -rf nv-codec-headers-master
+
+# Build ffmpeg with nvmpi patch
+wget -q https://ffmpeg.org/releases/ffmpeg-6.0.tar.xz
+tar xaf ffmpeg-*.tar.xz && rm ffmpeg-*.tar.xz && cd ffmpeg-*
+patch -p1 < ../jetson-ffmpeg-master/ffmpeg_patches/ffmpeg6.0_nvmpi.patch
+export PKG_CONFIG_PATH=$INSTALL_PREFIX/lib/pkgconfig
+# enable Jetson codecs but disable dGPU codecs
+./configure --cc='ccache gcc' --cxx='ccache g++' \
+            --enable-shared --disable-static --prefix=$INSTALL_PREFIX \
+            --enable-gpl --enable-libx264  --enable-libx265 \
+            --enable-nvmpi --enable-ffnvcodec --enable-cuda-llvm \
+            --disable-cuvid --disable-nvenc --disable-nvdec \
+    || { cat ffbuild/config.log && false; }
+make -j$(nproc)
+make install
+cd ../
+
+rm -rf /var/lib/apt/lists/*
+popd

docker/tensorrt/detector/build_python_tensorrt.sh

@@ -0,0 +1,28 @@
+#!/bin/bash
+
+set -euxo pipefail
+
+mkdir -p /trt-wheels
+
+if [[ "${TARGETARCH}" == "arm64" ]]; then
+
+  # NVIDIA supplies python-tensorrt for python3.8, but frigate uses python3.9,
+  # so we must build python-tensorrt ourselves.
+
+  # Get python-tensorrt source
+  mkdir /workspace
+  cd /workspace
+  git clone -b ${TENSORRT_VER} https://github.com/NVIDIA/TensorRT.git --depth=1
+
+  # Collect dependencies
+  EXT_PATH=/workspace/external && mkdir -p $EXT_PATH
+  pip3 install pybind11 && ln -s /usr/local/lib/python3.9/dist-packages/pybind11 $EXT_PATH/pybind11
+  ln -s /usr/include/python3.9 $EXT_PATH/python3.9
+  ln -s /usr/include/aarch64-linux-gnu/NvOnnxParser.h /workspace/TensorRT/parsers/onnx/
+
+  # Build wheel
+  cd /workspace/TensorRT/python
+  EXT_PATH=$EXT_PATH PYTHON_MAJOR_VERSION=3 PYTHON_MINOR_VERSION=9 TARGET_ARCHITECTURE=aarch64 /bin/bash ./build.sh
+  mv build/dist/*.whl /trt-wheels/
+
+fi

docker/tensorrt/detector/rootfs/etc/s6-overlay/s6-rc.d/trt-model-prepare/run

@@ -2,15 +2,21 @@
 # shellcheck shell=bash
 # Generate models for the TensorRT detector
 
+# One or more comma-separated models may be specified via the YOLO_MODELS env.
+# Append "-dla" to the model name to generate a DLA model with GPU fallback;
+# otherwise a GPU-only model will be generated.
+
 set -o errexit -o nounset -o pipefail
 
 MODEL_CACHE_DIR=${MODEL_CACHE_DIR:-"/config/model_cache/tensorrt"}
+TRT_VER=${TRT_VER:-$(cat /etc/TENSORRT_VER)}
 OUTPUT_FOLDER="${MODEL_CACHE_DIR}/${TRT_VER}"
 
 # Create output folder
 mkdir -p ${OUTPUT_FOLDER}
 
 FIRST_MODEL=true
+MODEL_DOWNLOAD=""
 MODEL_CONVERT=""
 
 for model in ${YOLO_MODELS//,/ }
@@ -20,9 +26,11 @@ do
 
     if [[ ! -f ${OUTPUT_FOLDER}/${model}.trt ]]; then
         if [[ ${FIRST_MODEL} = true ]]; then
+            MODEL_DOWNLOAD="${model%-dla}";
             MODEL_CONVERT="${model}"
             FIRST_MODEL=false;
         else
+            MODEL_DOWNLOAD+=",${model%-dla}";
             MODEL_CONVERT+=",${model}";
         fi
     else
@@ -35,19 +43,49 @@ if [[ -z ${MODEL_CONVERT} ]]; then
     exit 0
 fi
 
+# On Jetpack 4.6, the nvidia container runtime will mount several host nvidia libraries into the
+# container which should not be present in the image - if they are, TRT model generation will
+# fail or produce invalid models. Thus we must request the user to install them on the host in
+# order to run libyolo here.
+# On Jetpack 5.0, these libraries are not mounted by the runtime and are supplied by the image.
+if [[ "$(arch)" == "aarch64" ]]; then
+    if [[ ! -e /usr/lib/aarch64-linux-gnu/tegra ]]; then
+        echo "ERROR: Container must be launched with nvidia runtime"
+        exit 1
+    elif [[ ! -e /usr/lib/aarch64-linux-gnu/libnvinfer.so.8 ||
+            ! -e /usr/lib/aarch64-linux-gnu/libnvinfer_plugin.so.8 ||
+            ! -e /usr/lib/aarch64-linux-gnu/libnvparsers.so.8 ||
+            ! -e /usr/lib/aarch64-linux-gnu/libnvonnxparser.so.8 ]]; then
+        echo "ERROR: Please run the following on the HOST:"
+        echo "  sudo apt install libnvinfer8 libnvinfer-plugin8 libnvparsers8 libnvonnxparsers8 nvidia-container"
+        exit 1
+    fi
+fi
+
 echo "Generating the following TRT Models: ${MODEL_CONVERT}"
 
 # Build trt engine
 cd /usr/local/src/tensorrt_demos/yolo
 
-# Download yolo weights
+echo "Downloading yolo weights"
-./download_yolo.sh $MODEL_CONVERT > /dev/null
+./download_yolo.sh $MODEL_DOWNLOAD 2> /dev/null
 
 for model in ${MODEL_CONVERT//,/ }
 do
-    echo "Converting ${model} model"
+    python3 yolo_to_onnx.py -m ${model%-dla} > /dev/null
-    python3 yolo_to_onnx.py -m ${model} > /dev/null
+
-    python3 onnx_to_tensorrt.py -m ${model} > /dev/null
+    echo -e "\nGenerating ${model}.trt. This may take a few minutes.\n"; start=$(date +%s)
-    cp ${model}.trt ${OUTPUT_FOLDER}/${model}.trt
+    if [[ $model == *-dla ]]; then
+        cmd="python3 onnx_to_tensorrt.py -m ${model%-dla} --dla_core 0"
+    else
+        cmd="python3 onnx_to_tensorrt.py -m ${model}"
+    fi
+    $cmd > /tmp/onnx_to_tensorrt.log || { cat /tmp/onnx_to_tensorrt.log && continue; }
+
+    mv ${model%-dla}.trt ${OUTPUT_FOLDER}/${model}.trt;
     ln -s ${OUTPUT_FOLDER}/${model}.trt ${MODEL_CACHE_DIR}/${model}.trt
+    echo "Generated ${model}.trt in $(($(date +%s)-start)) seconds"
 done
+
+echo "Available tensorrt models:"
+cd ${OUTPUT_FOLDER} && ls *.trt;

docker/tensorrt/detector/tensorrt_libyolo.sh

@@ -8,7 +8,10 @@ SCRIPT_DIR="/usr/local/src/tensorrt_demos"
 git clone --depth 1 https://github.com/NateMeyer/tensorrt_demos.git -b conditional_download
 
 # Build libyolo
-cd ./tensorrt_demos/plugins && make all
+if [ ! -e /usr/local/cuda ]; then
+    ln -s /usr/local/cuda-* /usr/local/cuda
+fi
+cd ./tensorrt_demos/plugins && make all -j$(nproc)
 cp libyolo_layer.so /usr/local/lib/libyolo_layer.so
 
 # Store yolo scripts for later conversion

docker/tensorrt/requirements-arm64.txt

@@ -0,0 +1 @@
+cuda-python == 11.7; platform_machine == 'aarch64'

docker/tensorrt/requirements-models-arm64.txt

@@ -0,0 +1,3 @@
+onnx == 1.9.0; platform_machine == 'aarch64'
+protobuf == 3.20.3; platform_machine == 'aarch64'
+numpy == 1.23.*; platform_machine == 'aarch64'

docker/tensorrt/trt.hcl

@@ -1,19 +1,47 @@
+variable "ARCH" {
+  default = "amd64"
+}
+variable "BASE_IMAGE" {
+  default = null
+}
+variable "SLIM_BASE" {
+  default = null
+}
+variable "TRT_BASE" {
+  default = null
+}
+
+target "_build_args" {
+  args = {
+    BASE_IMAGE = BASE_IMAGE,
+    SLIM_BASE = SLIM_BASE,
+    TRT_BASE = TRT_BASE
+  }
+  platforms = ["linux/${ARCH}"]
+}
+
+target wget {
+  dockerfile = "docker/main/Dockerfile"
+  target = "wget"
+  inherits = ["_build_args"]
+}
+
 target deps {
   dockerfile = "docker/main/Dockerfile"
-  platforms = ["linux/amd64"]
   target = "deps"
+  inherits = ["_build_args"]
 }
 
 target rootfs {
   dockerfile = "docker/main/Dockerfile"
-  platforms = ["linux/amd64"]
   target = "rootfs"
+  inherits = ["_build_args"]
 }
 
 target wheels {
   dockerfile = "docker/main/Dockerfile"
-  platforms = ["linux/amd64"]
   target = "wheels"
+  inherits = ["_build_args"]
 }
 
 target devcontainer {
@@ -22,25 +50,43 @@ target devcontainer {
   target = "devcontainer"
 }
 
-target tensorrt {
+target "trt-deps" {
-  dockerfile = "docker/tensorrt/Dockerfile"
+  dockerfile = "docker/tensorrt/Dockerfile.base"
   context = "."
   contexts = {
     deps = "target:deps",
+  }
+  inherits = ["_build_args"]
+}
+
+target "tensorrt-base" {
+  dockerfile = "docker/tensorrt/Dockerfile.base"
+  context = "."
+  contexts = {
+    deps = "target:deps",
+  }
+  inherits = ["_build_args"]
+}
+
+target "tensorrt" {
+  dockerfile = "docker/tensorrt/Dockerfile.${ARCH}"
+  context = "."
+  contexts = {
+    wget = "target:wget",
+    tensorrt-base = "target:tensorrt-base",
     rootfs = "target:rootfs"
     wheels = "target:wheels"
   }
-  platforms = ["linux/amd64"]
   target = "frigate-tensorrt"
+  inherits = ["_build_args"]
 }
 
-target devcontainer-trt {
+target "devcontainer-trt" {
-  dockerfile = "docker/tensorrt/Dockerfile"
+  dockerfile = "docker/tensorrt/Dockerfile.amd64"
   context = "."
   contexts = {
-    deps = "target:deps",
+    wheels = "target:wheels",
-    rootfs = "target:rootfs"
+    trt-deps = "target:trt-deps",
-    wheels = "target:wheels"
     devcontainer = "target:devcontainer"
   }
   platforms = ["linux/amd64"]

docker/tensorrt/trt.mk

@@ -1,10 +1,26 @@
 BOARDS += trt
 
+JETPACK4_BASE ?= timongentzsch/l4t-ubuntu20-opencv:latest	# L4T 32.7.1 JetPack 4.6.1
+JETPACK5_BASE ?= nvcr.io/nvidia/l4t-tensorrt:r8.5.2-runtime	# L4T 35.3.1 JetPack 5.1.1
+X86_DGPU_ARGS := ARCH=amd64
+JETPACK4_ARGS := ARCH=arm64 BASE_IMAGE=$(JETPACK4_BASE) SLIM_BASE=$(JETPACK4_BASE) TRT_BASE=$(JETPACK4_BASE)
+JETPACK5_ARGS := ARCH=arm64 BASE_IMAGE=$(JETPACK5_BASE) SLIM_BASE=$(JETPACK5_BASE) TRT_BASE=$(JETPACK5_BASE)
+
 local-trt: version
-	docker buildx bake --load --file=docker/tensorrt/trt.hcl --set tensorrt.tags=frigate:latest-tensorrt tensorrt
+	$(X86_DGPU_ARGS) docker buildx bake --load --file=docker/tensorrt/trt.hcl --set tensorrt.tags=frigate:latest-tensorrt tensorrt
+
+local-trt-jp4: version
+	$(JETPACK4_ARGS) docker buildx bake --load --file=docker/tensorrt/trt.hcl --set tensorrt.tags=frigate:latest-tensorrt-jp4 tensorrt
+
+local-trt-jp5: version
+	$(JETPACK5_ARGS) docker buildx bake --load --file=docker/tensorrt/trt.hcl --set tensorrt.tags=frigate:latest-tensorrt-jp5 tensorrt
 
 build-trt:
-	docker buildx bake --file=docker/tensorrt/trt.hcl --set tensorrt.tags=$(IMAGE_REPO):${GITHUB_REF_NAME}-$(COMMIT_HASH)-tensorrt tensorrt
+	$(X86_DGPU_ARGS) docker buildx bake --file=docker/tensorrt/trt.hcl --set tensorrt.tags=$(IMAGE_REPO):${GITHUB_REF_NAME}-$(COMMIT_HASH)-tensorrt tensorrt
+	$(JETPACK4_ARGS) docker buildx bake --file=docker/tensorrt/trt.hcl --set tensorrt.tags=$(IMAGE_REPO):${GITHUB_REF_NAME}-$(COMMIT_HASH)-tensorrt-jp4 tensorrt
+	$(JETPACK5_ARGS) docker buildx bake --file=docker/tensorrt/trt.hcl --set tensorrt.tags=$(IMAGE_REPO):${GITHUB_REF_NAME}-$(COMMIT_HASH)-tensorrt-jp5 tensorrt
 
 push-trt: build-trt
-	docker buildx bake --push --file=docker/tensorrt/trt.hcl --set tensorrt.tags=$(IMAGE_REPO):${GITHUB_REF_NAME}-$(COMMIT_HASH)-tensorrt tensorrt
+	$(X86_DGPU_ARGS) docker buildx bake --push --file=docker/tensorrt/trt.hcl --set tensorrt.tags=$(IMAGE_REPO):${GITHUB_REF_NAME}-$(COMMIT_HASH)-tensorrt tensorrt
+	$(JETPACK4_ARGS) docker buildx bake --push --file=docker/tensorrt/trt.hcl --set tensorrt.tags=$(IMAGE_REPO):${GITHUB_REF_NAME}-$(COMMIT_HASH)-tensorrt-jp4 tensorrt
+	$(JETPACK5_ARGS) docker buildx bake --push --file=docker/tensorrt/trt.hcl --set tensorrt.tags=$(IMAGE_REPO):${GITHUB_REF_NAME}-$(COMMIT_HASH)-tensorrt-jp5 tensorrt

docs/docs/configuration/ffmpeg_presets.md

@@ -11,16 +11,18 @@ It is highly recommended to use hwaccel presets in the config. These presets not
 
 See [the hwaccel docs](/configuration/hardware_acceleration.md) for more info on how to setup hwaccel for your GPU / iGPU.
 
-| Preset                | Usage                        | Other Notes                                           |
+| Preset                | Usage                          | Other Notes                                           |
-| --------------------- | ---------------------------- | ----------------------------------------------------- |
+| --------------------- | ------------------------------ | ----------------------------------------------------- |
-| preset-rpi-32-h264    | 32 bit Rpi with h264 stream  |                                                       |
+| preset-rpi-32-h264    | 32 bit Rpi with h264 stream    |                                                       |
-| preset-rpi-64-h264    | 64 bit Rpi with h264 stream  |                                                       |
+| preset-rpi-64-h264    | 64 bit Rpi with h264 stream    |                                                       |
-| preset-vaapi          | Intel & AMD VAAPI            | Check hwaccel docs to ensure correct driver is chosen |
+| preset-vaapi          | Intel & AMD VAAPI              | Check hwaccel docs to ensure correct driver is chosen |
-| preset-intel-qsv-h264 | Intel QSV with h264 stream   | If issues occur recommend using vaapi preset instead  |
+| preset-intel-qsv-h264 | Intel QSV with h264 stream     | If issues occur recommend using vaapi preset instead  |
-| preset-intel-qsv-h265 | Intel QSV with h265 stream   | If issues occur recommend using vaapi preset instead  |
+| preset-intel-qsv-h265 | Intel QSV with h265 stream     | If issues occur recommend using vaapi preset instead  |
-| preset-nvidia-h264    | Nvidia GPU with h264 stream  |                                                       |
+| preset-nvidia-h264    | Nvidia GPU with h264 stream    |                                                       |
-| preset-nvidia-h265    | Nvidia GPU with h265 stream  |                                                       |
+| preset-nvidia-h265    | Nvidia GPU with h265 stream    |                                                       |
-| preset-nvidia-mjpeg   | Nvidia GPU with mjpeg stream | Recommend restreaming mjpeg and using nvidia-h264     |
+| preset-nvidia-mjpeg   | Nvidia GPU with mjpeg stream   | Recommend restreaming mjpeg and using nvidia-h264     |
+| preset-jetson-h264    | Nvidia Jetson with h264 stream |                                                       |
+| preset-jetson-h265    | Nvidia Jetson with h265 stream |                                                       |
 
 ### Input Args Presets
 

docs/docs/configuration/hardware_acceleration.md

@@ -246,3 +246,77 @@ If you do not see these processes, check the `docker logs` for the container and
 These instructions were originally based on the [Jellyfin documentation](https://jellyfin.org/docs/general/administration/hardware-acceleration.html#nvidia-hardware-acceleration-on-docker-linux).
 
 # Community Supported
+
+## NVIDIA Jetson (Orin AGX, Orin NX, Orin Nano*, Xavier AGX, Xavier NX, TX2, TX1, Nano)
+
+A separate set of docker images is available that is based on Jetpack/L4T. They comes with an `ffmpeg` build
+with codecs that use the Jetson's dedicated media engine. If your Jetson host is running Jetpack 4.6, use the
+`frigate-tensorrt-jp4` image, or if your Jetson host is running Jetpack 5.0+, use the `frigate-tensorrt-jp5`
+image. Note that the Orin Nano has no video encoder, so frigate will use software encoding on this platform,
+but the image will still allow hardware decoding and tensorrt object detection.
+
+You will need to use the image with the nvidia container runtime:
+
+### Docker Run CLI - Jetson
+
+```bash
+docker run -d \
+  ...
+  --runtime nvidia
+  ghcr.io/blakeblackshear/frigate-tensorrt-jp5
+```
+
+### Docker Compose - Jetson
+
+```yaml
+version: '2.4'
+services:
+  frigate:
+    ...
+    image: ghcr.io/blakeblackshear/frigate-tensorrt-jp5
+    runtime: nvidia   # Add this
+```
+
+:::note
+
+The `runtime:` tag is not supported on older versions of docker-compose. If you run into this, you can instead use the nvidia runtime system-wide by adding `"default-runtime": "nvidia"` to `/etc/docker/daemon.json`:
+
+```
+{
+    "runtimes": {
+        "nvidia": {
+            "path": "nvidia-container-runtime",
+            "runtimeArgs": []
+        }
+    },
+    "default-runtime": "nvidia"
+}
+```
+
+:::
+
+### Setup Decoder
+
+The decoder you need to pass in the `hwaccel_args` will depend on the input video.
+
+A list of supported codecs (you can use `ffmpeg -decoders | grep nvmpi` in the container to get the ones your card supports)
+
+```
+ V..... h264_nvmpi           h264 (nvmpi) (codec h264)
+ V..... hevc_nvmpi           hevc (nvmpi) (codec hevc)
+ V..... mpeg2_nvmpi          mpeg2 (nvmpi) (codec mpeg2video)
+ V..... mpeg4_nvmpi          mpeg4 (nvmpi) (codec mpeg4)
+ V..... vp8_nvmpi            vp8 (nvmpi) (codec vp8)
+ V..... vp9_nvmpi            vp9 (nvmpi) (codec vp9)
+```
+
+For example, for H264 video, you'll select `preset-jetson-h264`.
+
+```yaml
+ffmpeg:
+  hwaccel_args: preset-jetson-h264
+```
+
+If everything is working correctly, you should see a significant reduction in ffmpeg CPU load and power consumption.
+Verify that hardware decoding is working by running `jtop` (`sudo pip3 install -U jetson-stats`), which should show
+that NVDEC/NVDEC1 are in use.

docs/docs/configuration/index.md

@@ -101,7 +101,7 @@ detectors:
   # Required: name of the detector
   detector_name:
     # Required: type of the detector
-    # Frigate provided types include 'cpu', 'edgetpu', and 'openvino' (default: shown below)
+    # Frigate provided types include 'cpu', 'edgetpu', 'openvino' and 'tensorrt' (default: shown below)
     # Additional detector types can also be plugged in.
     # Detectors may require additional configuration.
     # Refer to the Detectors configuration page for more information.
@@ -414,6 +414,8 @@ snapshots:
     # Optional: Per object retention days
     objects:
       person: 15
+  # Optional: quality of the encoded jpeg, 0-100 (default: shown below)
+  quality: 70
 
 # Optional: RTMP configuration
 # NOTE: RTMP is deprecated in favor of restream

docs/docs/configuration/object_detectors.md

@@ -196,7 +196,9 @@ The model used for TensorRT must be preprocessed on the same hardware platform t
 
 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.
 
-To by default, the `yolov7-tiny-416` model 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.  To select no model generation, set the variable to an empty string, `YOLO_MODELS=""`.  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.
+By default, the `yolov7-320` model 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. To select no model generation, set the variable to an empty string, `YOLO_MODELS=""`. 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.
 
@@ -252,11 +254,11 @@ detectors:
     device: 0 #This is the default, select the first GPU
 
 model:
-  path: /config/model_cache/tensorrt/yolov7-tiny-416.trt
+  path: /config/model_cache/tensorrt/yolov7-320.trt
   input_tensor: nchw
   input_pixel_format: rgb
-  width: 416
+  width: 320
-  height: 416
+  height: 320
 ```
 
 ## Deepstack / CodeProject.AI Server Detector

docs/docs/development/contributing.md

@@ -101,12 +101,18 @@ This should show <50% CPU in top, and ~80% CPU without `-c:v h264_v4l2m2m`.
 ffmpeg -c:v h264_v4l2m2m -re -stream_loop -1 -i https://streams.videolan.org/ffmpeg/incoming/720p60.mp4 -f rawvideo -pix_fmt yuv420p pipe: > /dev/null
 ```
 
-**NVIDIA**
+**NVIDIA GPU**
 
 ```shell
 ffmpeg -c:v h264_cuvid -re -stream_loop -1 -i https://streams.videolan.org/ffmpeg/incoming/720p60.mp4 -f rawvideo -pix_fmt yuv420p pipe: > /dev/null
 ```
 
+**NVIDIA Jetson**
+
+```shell
+ffmpeg -c:v h264_nvmpi -re -stream_loop -1 -i https://streams.videolan.org/ffmpeg/incoming/720p60.mp4 -f rawvideo -pix_fmt yuv420p pipe: > /dev/null
+```
+
 **VAAPI**
 
 ```shell

docs/docs/frigate/hardware.md

@@ -70,7 +70,7 @@ Inference speeds vary greatly depending on the CPU, GPU, or VPU used, some known
 | Intel i5 1135G7      | 10 - 15 ms      |                                                                       |
 | Intel i5 12600K      | ~ 15 ms         | Inference speeds on CPU were ~ 35 ms                                  |
 
-### TensorRT
+### TensorRT - Nvidia GPU
 
 The TensortRT detector is able to run on x86 hosts that have an Nvidia GPU which supports the 12.x series of CUDA libraries. The minimum driver version on the host system must be `>=525.60.13`. Also the GPU must support a Compute Capability of `5.0` or greater. This generally correlates to a Maxwell-era GPU or newer, check the [TensorRT docs for more info](/configuration/object_detectors#nvidia-tensorrt-detector).
 
@@ -87,6 +87,14 @@ Inference speeds will vary greatly depending on the GPU and the model used.
 | Quadro P400 2GB | 20 - 25 ms      |
 | Quadro P2000    | ~ 12 ms         |
 
+### Community Supported:
+
+#### Nvidia Jetson
+
+Frigate supports all Jetson boards, from the inexpensive Jetson Nano to the powerful Jetson Orin AGX. It will [make use of the Jetson's hardware media engine](/configuration/hardware_acceleration#nvidia-jetson-orin-agx-orin-nx-orin-nano-xavier-agx-xavier-nx-tx2-tx1-nano) when configured with the [appropriate presets](/configuration/ffmpeg_presets#hwaccel-presets), and will make use of the Jetson's GPU and DLA for object detection when configured with the [TensorRT detector](/configuration/object_detectors#nvidia-tensorrt-detector).
+
+Inference speed will vary depending on the YOLO model, jetson platform and jetson nvpmodel (GPU/DLA/EMC clock speed). It is typically 20-40 ms for most models. The DLA is more efficient than the GPU, but not faster, so using the DLA will reduce power consumption but will slightly increase inference time.
+
 ## What does Frigate use the CPU for and what does it use a detector for? (ELI5 Version)
 
 This is taken from a [user question on reddit](https://www.reddit.com/r/homeassistant/comments/q8mgau/comment/hgqbxh5/?utm_source=share&utm_medium=web2x&context=3). Modified slightly for clarity.

docs/docs/frigate/installation.md

@@ -93,6 +93,9 @@ The following officially supported builds are available:
 
 The following community supported builds are available:
 
+`ghcr.io/blakeblackshear/frigate:stable-tensorrt-jp5` - Frigate build optimized for nvidia Jetson devices running Jetpack 5
+`ghcr.io/blakeblackshear/frigate:stable-tensorrt-jp4` - Frigate build optimized for nvidia Jetson devices running Jetpack 4.6
+
 :::
 
 ```yaml

frigate/config.py

@@ -804,9 +804,19 @@ class CameraConfig(FrigateBaseModel):
             ffmpeg_input.global_args or self.ffmpeg.global_args
         )
         hwaccel_args = get_ffmpeg_arg_list(
-            parse_preset_hardware_acceleration_decode(ffmpeg_input.hwaccel_args)
+            parse_preset_hardware_acceleration_decode(
+                ffmpeg_input.hwaccel_args,
+                self.detect.fps,
+                self.detect.width,
+                self.detect.height,
+            )
             or ffmpeg_input.hwaccel_args
-            or parse_preset_hardware_acceleration_decode(self.ffmpeg.hwaccel_args)
+            or parse_preset_hardware_acceleration_decode(
+                self.ffmpeg.hwaccel_args,
+                self.detect.fps,
+                self.detect.width,
+                self.detect.height,
+            )
             or self.ffmpeg.hwaccel_args
         )
         input_args = get_ffmpeg_arg_list(

frigate/ffmpeg_presets.py

@@ -55,58 +55,16 @@ _user_agent_args = [
 ]
 
 PRESETS_HW_ACCEL_DECODE = {
-    "preset-rpi-32-h264": ["-c:v:1", "h264_v4l2m2m"],
+    "preset-rpi-32-h264": "-c:v:1 h264_v4l2m2m",
-    "preset-rpi-64-h264": ["-c:v:1", "h264_v4l2m2m"],
+    "preset-rpi-64-h264": "-c:v:1 h264_v4l2m2m",
-    "preset-vaapi": [
+    "preset-vaapi": f"-hwaccel_flags allow_profile_mismatch -hwaccel vaapi -hwaccel_devic {_gpu_selector.get_selected_gpu()} -hwaccel_output_format vaapi",
-        "-hwaccel_flags",
+    "preset-intel-qsv-h264": f"-hwaccel qsv -qsv_devic {_gpu_selector.get_selected_gpu()} -hwaccel_output_format qsv -c:v h264_qsv",
-        "allow_profile_mismatch",
+    "preset-intel-qsv-h265": f"-load_plugin hevc_hw -hwaccel qsv -qsv_device {_gpu_selector.get_selected_gpu()} -hwaccel_output_format qsv -c:v hevc_qsv",
-        "-hwaccel",
+    "preset-nvidia-h264": "-hwaccel cuda -hwaccel_output_format cuda",
-        "vaapi",
+    "preset-nvidia-h265": "-hwaccel cuda -hwaccel_output_format cuda",
-        "-hwaccel_device",
+    "preset-nvidia-mjpeg": "-hwaccel cuda -hwaccel_output_format cuda",
-        _gpu_selector.get_selected_gpu(),
+    "preset-jetson-h264": "-c:v h264_nvmpi -resize {1}x{2}",
-        "-hwaccel_output_format",
+    "preset-jetson-h265": "-c:v hevc_nvmpi -resize {1}x{2}",
-        "vaapi",
-    ],
-    "preset-intel-qsv-h264": [
-        "-hwaccel",
-        "qsv",
-        "-qsv_device",
-        _gpu_selector.get_selected_gpu(),
-        "-hwaccel_output_format",
-        "qsv",
-        "-c:v",
-        "h264_qsv",
-    ],
-    "preset-intel-qsv-h265": [
-        "-load_plugin",
-        "hevc_hw",
-        "-hwaccel",
-        "qsv",
-        "-qsv_device",
-        _gpu_selector.get_selected_gpu(),
-        "-hwaccel_output_format",
-        "qsv",
-        "-c:v",
-        "hevc_qsv",
-    ],
-    "preset-nvidia-h264": [
-        "-hwaccel",
-        "cuda",
-        "-hwaccel_output_format",
-        "cuda",
-    ],
-    "preset-nvidia-h265": [
-        "-hwaccel",
-        "cuda",
-        "-hwaccel_output_format",
-        "cuda",
-    ],
-    "preset-nvidia-mjpeg": [
-        "-hwaccel",
-        "cuda",
-        "-hwaccel_output_format",
-        "cuda",
-    ],
 }
 
 PRESETS_HW_ACCEL_SCALE = {
@@ -117,6 +75,8 @@ PRESETS_HW_ACCEL_SCALE = {
     "preset-intel-qsv-h265": "-r {0} -vf vpp_qsv=framerate={0}:w={1}:h={2}:format=nv12,hwdownload,format=nv12,format=yuv420p",
     "preset-nvidia-h264": "-r {0} -vf fps={0},scale_cuda=w={1}:h={2}:format=nv12,hwdownload,format=nv12,format=yuv420p",
     "preset-nvidia-h265": "-r {0} -vf fps={0},scale_cuda=w={1}:h={2}:format=nv12,hwdownload,format=nv12,format=yuv420p",
+    "preset-jetson-h264": "-r {0}",  # scaled in decoder
+    "preset-jetson-h265": "-r {0}",  # scaled in decoder
     "default": "-r {0} -vf fps={0},scale={1}:{2}",
 }
 
@@ -128,6 +88,8 @@ PRESETS_HW_ACCEL_ENCODE_BIRDSEYE = {
     "preset-intel-qsv-h265": "ffmpeg -hide_banner {0} -c:v h264_qsv -g 50 -bf 0 -profile:v high -level:v 4.1 -async_depth:v 1 {1}",
     "preset-nvidia-h264": "ffmpeg -hide_banner {0} -c:v h264_nvenc -g 50 -profile:v high -level:v auto -preset:v p2 -tune:v ll {1}",
     "preset-nvidia-h265": "ffmpeg -hide_banner {0} -c:v h264_nvenc -g 50 -profile:v high -level:v auto -preset:v p2 -tune:v ll {1}",
+    "preset-jetson-h264": "ffmpeg -hide_banner {0} -c:v h264_nvmpi -profile high {1}",
+    "preset-jetson-h265": "ffmpeg -hide_banner {0} -c:v h264_nvmpi -profile high {1}",
     "default": "ffmpeg -hide_banner {0} -c:v libx264 -g 50 -profile:v high -level:v 4.1 -preset:v superfast -tune:v zerolatency {1}",
 }
 
@@ -139,16 +101,28 @@ PRESETS_HW_ACCEL_ENCODE_TIMELAPSE = {
     "preset-intel-qsv-h265": "ffmpeg -hide_banner {0} -c:v hevc_qsv -g 50 -bf 0 -profile:v high -level:v 4.1 -async_depth:v 1 {1}",
     "preset-nvidia-h264": "ffmpeg -hide_banner -hwaccel cuda -hwaccel_output_format cuda -extra_hw_frames 8 {0} -c:v h264_nvenc {1}",
     "preset-nvidia-h265": "ffmpeg -hide_banner -hwaccel cuda -hwaccel_output_format cuda -extra_hw_frames 8 {0} -c:v hevc_nvenc {1}",
+    "preset-jetson-h264": "ffmpeg -hide_banner {0} -c:v h264_nvmpi -profile high {1}",
+    "preset-jetson-h265": "ffmpeg -hide_banner {0} -c:v hevc_nvmpi -profile high {1}",
     "default": "ffmpeg -hide_banner {0} -c:v libx264 -preset:v ultrafast -tune:v zerolatency {1}",
 }
 
 
-def parse_preset_hardware_acceleration_decode(arg: Any) -> list[str]:
+def parse_preset_hardware_acceleration_decode(
+    arg: Any,
+    fps: int,
+    width: int,
+    height: int,
+) -> list[str]:
     """Return the correct preset if in preset format otherwise return None."""
     if not isinstance(arg, str):
         return None
 
-    return PRESETS_HW_ACCEL_DECODE.get(arg, None)
+    decode = PRESETS_HW_ACCEL_DECODE.get(arg, None)
+
+    if not decode:
+        return None
+
+    return decode.format(fps, width, height).split(" ")
 
 
 def parse_preset_hardware_acceleration_scale(
@@ -160,20 +134,13 @@ def parse_preset_hardware_acceleration_scale(
 ) -> list[str]:
     """Return the correct scaling preset or default preset if none is set."""
     if not isinstance(arg, str) or " " in arg:
-        scale = PRESETS_HW_ACCEL_SCALE["default"].format(fps, width, height).split(" ")
+        scale = PRESETS_HW_ACCEL_SCALE["default"]
-        scale.extend(detect_args)
-        return scale
-
-    scale = PRESETS_HW_ACCEL_SCALE.get(arg, "")
-
-    if scale:
-        scale = scale.format(fps, width, height).split(" ")
-        scale.extend(detect_args)
-        return scale
     else:
-        scale = scale.format(fps, width, height).split(" ")
+        scale = PRESETS_HW_ACCEL_SCALE.get(arg, "")
-        scale.extend(detect_args)
+
-        return scale
+    scale = scale.format(fps, width, height).split(" ")
+    scale.extend(detect_args)
+    return scale
 
 
 class EncodeTypeEnum(str, Enum):
@@ -193,6 +160,10 @@ def parse_preset_hardware_acceleration_encode(
     if not isinstance(arg, str):
         return arg_map["default"].format(input, output)
 
+    # Not all jetsons have HW encoders, so fall back to default SW encoder if not
+    if arg.startswith("preset-jetson-") and not os.path.exists("/dev/nvhost-msenc"):
+        arg = "default"
+
     return arg_map.get(arg, arg_map["default"]).format(
         input,
         output,

frigate/stats.py

@@ -22,6 +22,7 @@ from frigate.util.services import (
     get_bandwidth_stats,
     get_cpu_stats,
     get_intel_gpu_stats,
+    get_jetson_stats,
     get_nvidia_gpu_stats,
 )
 from frigate.version import VERSION
@@ -180,6 +181,15 @@ async def set_gpu_stats(
             else:
                 stats["nvidia-gpu"] = {"gpu": -1, "mem": -1}
                 hwaccel_errors.append(args)
+        elif "nvmpi" in args or "jetson" in args:
+            # nvidia Jetson
+            jetson_usage = get_jetson_stats()
+
+            if jetson_usage:
+                stats["jetson-gpu"] = jetson_usage
+            else:
+                stats["jetson-gpu"] = {"gpu": -1, "mem": -1}
+                hwaccel_errors.append(args)
         elif "qsv" in args:
             if not config.telemetry.stats.intel_gpu_stats:
                 continue

frigate/util/services.py

@@ -309,6 +309,21 @@ def get_nvidia_gpu_stats() -> dict[int, dict]:
         return results
 
 
+def get_jetson_stats() -> dict[int, dict]:
+    results = {}
+
+    try:
+        results["mem"] = "-"  # no discrete gpu memory
+
+        with open("/sys/devices/gpu.0/load", "r") as f:
+            gpuload = float(f.readline()) / 10
+            results["gpu"] = f"{gpuload}%"
+    except Exception:
+        return None
+
+    return results
+
+
 def ffprobe_stream(path: str) -> sp.CompletedProcess:
     """Run ffprobe on stream."""
     clean_path = escape_special_characters(path)