Back to Details

frigate-configurator

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Frigate NVR Configuration Expert

Frigate NVR配置专家

Comprehensive Frigate NVR configuration assistance with optimized YAML generation, detector setup, and troubleshooting.
提供全面的Frigate NVR配置支持，包括优化的YAML生成、检测器设置和故障排查。

BEFORE YOU START

开始之前

This skill prevents 12+ common errors and saves ~60% tokens on Frigate configuration.
MetricWithout SkillWith Skill
Setup Time2-4 hours30-45 min
Common Errors12+0
Token Usage~15,000~6,000
本技能可避免12种以上常见错误，并能减少约60%的Frigate配置token消耗。
指标未使用本技能使用本技能
部署时间2-4小时30-45分钟
常见错误12+0
Token消耗~15,000~6,000

Known Issues This Skill Prevents

本技能可避免的已知问题

  1. Bus errors from insufficient shared memory allocation
  2. Green/distorted video from incorrect resolution configuration
  3. Database locked errors when using network storage for SQLite
  4. Missing audio in recordings due to default audio stripping
  5. MQTT connection failures from using localhost in Docker
  6. Coral TPU not detected due to missing device passthrough
  7. High CPU usage from missing hardware acceleration
  8. False positives from missing motion masks on timestamps
  9. No alerts triggered due to misconfigured required_zones
  10. Recording corruption from h265 streams without transcoding
  11. go2rtc WebRTC failures from missing STUN configuration
  12. Object detection misses from wrong detect stream resolution
  1. 共享内存分配不足导致的总线错误
  2. 分辨率配置错误导致的画面绿屏/失真
  3. SQLite数据库存储在网络存储时出现的数据库锁定错误
  4. 默认音频剥离导致的录像无音频问题
  5. Docker容器内使用localhost导致的MQTT连接失败
  6. 未进行设备透传导致Coral TPU无法被识别
  7. 未启用硬件加速导致的高CPU占用
  8. 未为时间戳添加运动遮罩导致的误报
  9. required_zones配置错误导致的无警报触发
  10. h265流未转码导致的录像损坏
  11. 未配置STUN导致的go2rtc WebRTC失败
  12. 检测流分辨率错误导致的目标检测遗漏

Quick Start

快速开始

Step 1: Create Minimal Configuration

步骤1：创建最简配置

yaml
mqtt:
  enabled: false

cameras:
  front_door:
    ffmpeg:
      inputs:
        - path: rtsp://user:pass@192.168.1.100:554/stream1
          roles:
            - detect
    detect:
      width: 1280
      height: 720
      fps: 5
Why this matters: Start with the absolute minimum to verify camera connectivity before adding complexity. Frigate requires explicit detect stream role assignment.
yaml
mqtt:
  enabled: false

cameras:
  front_door:
    ffmpeg:
      inputs:
        - path: rtsp://user:pass@192.168.1.100:554/stream1
          roles:
            - detect
    detect:
      width: 1280
      height: 720
      fps: 5
为什么这很重要： 在添加复杂配置前，先从最简配置开始验证摄像头连接性。Frigate需要明确指定检测流的角色。

Step 2: Add Hardware-Accelerated Detector

步骤2：添加硬件加速检测器

yaml
detectors:
  coral:
    type: edgetpu
    device: usb
yaml
detectors:
  coral:
    type: edgetpu
    device: usb

OR for Intel with OpenVINO:

或者针对Intel平台使用OpenVINO:

detectors: ov: type: openvino device: GPU

**Why this matters:** CPU detection is not recommended for production. Even a single USB Coral TPU dramatically reduces CPU usage and improves detection latency.
detectors: ov: type: openvino device: GPU

**为什么这很重要：** 不推荐在生产环境中使用CPU进行检测。即使是一个USB Coral TPU也能显著降低CPU占用并提升检测延迟。

Step 3: Enable Recording with Retention

步骤3：启用带保留策略的录像功能

yaml
record:
  enabled: true
  retain:
    days: 1
    mode: motion
  alerts:
    retain:
      days: 14
  detections:
    retain:
      days: 7

cameras:
  front_door:
    ffmpeg:
      inputs:
        - path: rtsp://user:pass@192.168.1.100:554/stream1
          roles:
            - detect
        - path: rtsp://user:pass@192.168.1.100:554/stream2
          roles:
            - record
Why this matters: Use separate streams for detect (low-res) and record (high-res) to optimize performance. Retention modes prevent storage from filling up.
yaml
record:
  enabled: true
  retain:
    days: 1
    mode: motion
  alerts:
    retain:
      days: 14
  detections:
    retain:
      days: 7

cameras:
  front_door:
    ffmpeg:
      inputs:
        - path: rtsp://user:pass@192.168.1.100:554/stream1
          roles:
            - detect
        - path: rtsp://user:pass@192.168.1.100:554/stream2
          roles:
            - record
为什么这很重要： 为检测（低分辨率）和录像（高分辨率）使用独立流以优化性能。保留策略可防止存储被占满。

Critical Rules

关键规则

Always Do

务必遵守

  • Use 
    width
    and 
    height
    that match your camera's ACTUAL resolution (verify with VLC)
  • Set 
    detect
    fps between 5-10 (higher wastes resources, lower misses events)
  • Use separate streams for 
    detect
    (sub-stream) and 
    record
    (main stream)
  • Allocate adequate 
    shm-size
    in Docker (64MB minimum per camera)
  • Create motion masks for timestamp overlays and areas with constant motion
  • Use environment variables for credentials: 
    {FRIGATE_RTSP_PASSWORD}
  • Test RTSP URLs in VLC first before adding to Frigate config
  • 使用与摄像头实际分辨率匹配的
    width
    和
    height
    （通过VLC验证）
  • 将
    detect
    帧率设置在5-10之间（过高浪费资源，过低会遗漏事件）
  • 为
    detect
    （子流）和
    record
    （主流）使用独立的流
  • 在Docker中分配足够的
    shm-size
    （每台摄像头至少64MB）
  • 为时间戳覆盖层和持续运动区域创建运动遮罩
  • 使用环境变量存储凭据：
    {FRIGATE_RTSP_PASSWORD}
  • 在添加到Frigate配置前，先在VLC中测试RTSP URL

Never Do

绝对禁止

  • Never use 
    localhost
    or 
    127.0.0.1
    for MQTT inside Docker containers
  • Never set detect resolution higher than 1280x720 (wastes detector capacity)
  • Never enable recording without specifying retention policy
  • Never mount 
    /media/frigate
    on network storage without relocating database
  • Never mix multiple detector types for object detection (e.g., Coral + OpenVINO)
  • Never use UDP RTSP transport without explicit configuration (TCP is default)
  • 不要在Docker容器内为MQTT使用
    localhost
    或
    127.0.0.1
  • 不要将检测分辨率设置为1280x720以上（浪费检测器容量）
  • 不要在未指定保留策略的情况下启用录像
  • 不要在未迁移数据库的情况下将
    /media/frigate
    挂载到网络存储
  • 不要混合使用多种检测器类型进行目标检测（如Coral + OpenVINO）
  • 不要在未显式配置的情况下使用UDP RTSP传输（默认是TCP）

Common Mistakes

常见错误示例

Wrong:
yaml
cameras:
  cam1:
    ffmpeg:
      inputs:
        - path: rtsp://192.168.1.100/stream
          roles:
            - detect
            - record
    detect:
      width: 1920
      height: 1080
      fps: 30
Correct:
yaml
cameras:
  cam1:
    ffmpeg:
      inputs:
        - path: rtsp://192.168.1.100/substream
          roles:
            - detect
        - path: rtsp://192.168.1.100/mainstream
          roles:
            - record
    detect:
      width: 1280
      height: 720
      fps: 5
Why: Using 1080p@30fps for detection wastes resources. Detection works best at 720p or lower at 5fps. Always use the camera's sub-stream for detection and main stream for recording.
错误配置：
yaml
cameras:
  cam1:
    ffmpeg:
      inputs:
        - path: rtsp://192.168.1.100/stream
          roles:
            - detect
            - record
    detect:
      width: 1920
      height: 1080
      fps: 30
正确配置：
yaml
cameras:
  cam1:
    ffmpeg:
      inputs:
        - path: rtsp://192.168.1.100/substream
          roles:
            - detect
        - path: rtsp://192.168.1.100/mainstream
          roles:
            - record
    detect:
      width: 1280
      height: 720
      fps: 5
原因： 使用1080p@30fps进行检测会浪费资源。检测在720p或更低分辨率、5帧率下效果最佳。务必使用摄像头的子流进行检测，主流进行录像。

Known Issues Prevention

已知问题预防方案

IssueRoot CauseSolution
Bus ErrorInsufficient shared memorySet 
shm-size: 256mb
in docker-compose
Database LockedSQLite on network storageUse 
database.path: /config/frigate.db
Green/Distorted VideoWrong resolution in configMatch camera's actual output resolution
No Audio in RecordingsDefault audio removalUse 
preset-record-generic-audio-aac
MQTT Connection Failedlocalhost in DockerUse host IP address instead
Coral Not DetectedMissing device passthroughAdd 
/dev/bus/usb
to Docker devices
High CPU UsageMissing hwaccelAdd appropriate preset (vaapi/qsv/nvidia)
Missing AlertsNo required_zonesConfigure zones with review.alerts.required_zones
UDP Stream FailuresTCP is default in FrigateAdd 
preset-rtsp-udp
to input args
问题根本原因解决方案
总线错误共享内存不足在docker-compose中设置
shm-size: 256mb
数据库锁定SQLite存储在网络存储使用
database.path: /config/frigate.db
画面绿屏/失真配置的分辨率错误匹配摄像头的实际输出分辨率
录像无音频默认移除音频使用
preset-record-generic-audio-aac
MQTT连接失败Docker内使用localhost使用主机IP地址替代
Coral TPU未被识别未进行设备透传在Docker设备映射中添加
/dev/bus/usb
CPU占用过高未启用硬件加速添加对应的预设（vaapi/qsv/nvidia）
无警报触发未配置required_zones通过review.alerts.required_zones配置区域
UDP流失败Frigate默认使用TCP在输入参数中添加
preset-rtsp-udp

Configuration Reference

配置参考

config.yml Structure

config.yml结构

yaml
undefined
yaml
undefined

MQTT Configuration (optional but recommended)

MQTT配置（可选但推荐）

mqtt: enabled: true host: 192.168.1.50 port: 1883 user: "{FRIGATE_MQTT_USER}" password: "{FRIGATE_MQTT_PASSWORD}"
mqtt: enabled: true host: 192.168.1.50 port: 1883 user: "{FRIGATE_MQTT_USER}" password: "{FRIGATE_MQTT_PASSWORD}"

Detector Configuration

检测器配置

detectors: coral: type: edgetpu device: usb # or pci for M.2/PCIe
detectors: coral: type: edgetpu device: usb # 或pci用于M.2/PCIe接口

Global Object Settings

全局目标设置

objects: track: - person - car - dog - cat filters: person: min_area: 5000 max_area: 100000 threshold: 0.7
objects: track: - person - car - dog - cat filters: person: min_area: 5000 max_area: 100000 threshold: 0.7

Recording Settings

录像设置

record: enabled: true retain: days: 1 mode: motion alerts: retain: days: 14 detections: retain: days: 7
record: enabled: true retain: days: 1 mode: motion alerts: retain: days: 14 detections: retain: days: 7

Snapshot Settings

快照设置

snapshots: enabled: true retain: default: 7
snapshots: enabled: true retain: default: 7

Camera Configuration

摄像头配置

cameras: front_door: enabled: true ffmpeg: inputs: - path: "rtsp://{FRIGATE_RTSP_USER}:{FRIGATE_RTSP_PASSWORD}@192.168.1.100:554/stream1" input_args: preset-rtsp-restream roles: - detect - path: "rtsp://{FRIGATE_RTSP_USER}:{FRIGATE_RTSP_PASSWORD}@192.168.1.100:554/stream0" input_args: preset-rtsp-restream roles: - record output_args: record: preset-record-generic-audio-aac detect: width: 1280 height: 720 fps: 5 motion: mask: - 0,0,200,0,200,100,0,100 # Timestamp area zones: front_yard: coordinates: 100,500,400,500,400,720,100,720 objects: - person - car review: alerts: required_zones: - front_yard

**Key settings:**
- `detect.fps`: 5 is optimal for most cameras (reduces detector load)
- `detect.width/height`: Must match actual camera sub-stream resolution
- `record.retain.mode`: Use `motion` or `active_objects` to save storage
- `motion.mask`: Define polygons as comma-separated coordinates
- `zones.coordinates`: Bottom-center of bounding box determines zone presence
cameras: front_door: enabled: true ffmpeg: inputs: - path: "rtsp://{FRIGATE_RTSP_USER}:{FRIGATE_RTSP_PASSWORD}@192.168.1.100:554/stream1" input_args: preset-rtsp-restream roles: - detect - path: "rtsp://{FRIGATE_RTSP_USER}:{FRIGATE_RTSP_PASSWORD}@192.168.1.100:554/stream0" input_args: preset-rtsp-restream roles: - record output_args: record: preset-record-generic-audio-aac detect: width: 1280 height: 720 fps: 5 motion: mask: - 0,0,200,0,200,100,0,100 # 时间戳区域 zones: front_yard: coordinates: 100,500,400,500,400,720,100,720 objects: - person - car review: alerts: required_zones: - front_yard

**关键设置：**
- `detect.fps`：对于大多数摄像头来说，5是最优值（可降低检测器负载）
- `detect.width/height`：必须与摄像头子流的实际分辨率匹配
- `record.retain.mode`：使用`motion`或`active_objects`以节省存储空间
- `motion.mask`：用逗号分隔的坐标定义多边形
- `zones.coordinates`：目标框的底部中心位置决定其所属区域

Hardware Acceleration Presets

硬件加速预设

Intel (6th Gen+)

Intel（第6代及以上）

yaml
undefined
yaml
undefined

For Intel gen8+ (prefer QSV)

适用于Intel第8代及以上平台（优先使用QSV）

ffmpeg: hwaccel_args: preset-intel-qsv-h264 # or preset-intel-qsv-h265
ffmpeg: hwaccel_args: preset-intel-qsv-h264 # 或preset-intel-qsv-h265

For Intel gen1-gen7 (use VAAPI)

适用于Intel第1-7代平台（使用VAAPI）

ffmpeg: hwaccel_args: preset-vaapi
undefined
ffmpeg: hwaccel_args: preset-vaapi
undefined

NVIDIA GPU

NVIDIA GPU

yaml
ffmpeg:
  hwaccel_args: preset-nvidia
Requires NVIDIA Container Toolkit:
yaml
undefined
yaml
ffmpeg:
  hwaccel_args: preset-nvidia
需要NVIDIA容器工具包：
yaml
undefined

docker-compose.yml

docker-compose.yml

services: frigate: runtime: nvidia environment: - NVIDIA_VISIBLE_DEVICES=all
undefined
services: frigate: runtime: nvidia environment: - NVIDIA_VISIBLE_DEVICES=all
undefined

AMD GPU

AMD GPU

yaml
ffmpeg:
  hwaccel_args: preset-vaapi
yaml
ffmpeg:
  hwaccel_args: preset-vaapi

docker-compose.yml

docker-compose.yml

environment:
  • LIBVA_DRIVER_NAME=radeonsi
undefined
environment:
  • LIBVA_DRIVER_NAME=radeonsi
undefined

Raspberry Pi

Raspberry Pi

yaml
undefined
yaml
undefined

Raspberry Pi 4/5

Raspberry Pi 4/5

ffmpeg: hwaccel_args: preset-rpi-64-h264 # or preset-rpi-64-h265

Requires: `gpu_mem=128` in `/boot/config.txt` and device mapping in Docker.
ffmpeg: hwaccel_args: preset-rpi-64-h264 # 或preset-rpi-64-h265

要求：在`/boot/config.txt`中设置`gpu_mem=128`，并在Docker中进行设备映射。

Object Detector Types

目标检测器类型

USB Coral TPU

USB Coral TPU

yaml
detectors:
  coral:
    type: edgetpu
    device: usb  # Single USB Coral
    # device: usb:0  # First of multiple USB Corals
Docker device mapping:
yaml
devices:
  - /dev/bus/usb:/dev/bus/usb
yaml
detectors:
  coral:
    type: edgetpu
    device: usb  # 单个USB Coral
    # device: usb:0  # 多个USB Coral中的第一个
Docker设备映射：
yaml
devices:
  - /dev/bus/usb:/dev/bus/usb

M.2/PCIe Coral TPU

M.2/PCIe Coral TPU

yaml
detectors:
  coral:
    type: edgetpu
    device: pci
    # device: pci:0  # First of multiple PCIe Corals
yaml
detectors:
  coral:
    type: edgetpu
    device: pci
    # device: pci:0  # 多个PCIe Coral中的第一个

OpenVINO (Intel)

OpenVINO（Intel）

yaml
detectors:
  ov:
    type: openvino
    device: GPU  # or CPU

model:
  path: /openvino-model/ssdlite_mobilenet_v2.xml
  width: 300
  height: 300
yaml
detectors:
  ov:
    type: openvino
    device: GPU  # 或CPU

model:
  path: /openvino-model/ssdlite_mobilenet_v2.xml
  width: 300
  height: 300

ONNX (Multi-GPU)

ONNX（多GPU）

yaml
detectors:
  onnx:
    type: onnx
    # Automatically uses: ROCm (AMD), OpenVINO (Intel), TensorRT (NVIDIA)
yaml
detectors:
  onnx:
    type: onnx
    # 自动适配：ROCm（AMD）、OpenVINO（Intel）、TensorRT（NVIDIA）

Advanced Features

高级功能

Zone-Based Speed Estimation

基于区域的速度估算

yaml
zones:
  driveway:
    coordinates: 100,500,400,500,400,720,100,720
    distances:
      - "100,500|400,500|20ft"  # 20 feet between points
    speed:
      threshold: 15  # Minimum mph to register
yaml
zones:
  driveway:
    coordinates: 100,500,400,500,400,720,100,720
    distances:
      - "100,500|400,500|20ft"  # 两点间距离为20英尺
    speed:
      threshold: 15  # 触发速度的最低阈值（英里/小时）

Audio Detection

音频检测

yaml
audio:
  enabled: true
  listen:
    - bark
    - fire_alarm
    - scream
    - speech

cameras:
  front_door:
    ffmpeg:
      inputs:
        - path: rtsp://camera/stream
          roles:
            - audio
yaml
audio:
  enabled: true
  listen:
    - bark
    - fire_alarm
    - scream
    - speech

cameras:
  front_door:
    ffmpeg:
      inputs:
        - path: rtsp://camera/stream
          roles:
            - audio

GenAI Event Descriptions

生成式AI事件描述

yaml
genai:
  enabled: true
  provider: ollama
  base_url: http://192.168.1.100:11434
  model: llava
yaml
genai:
  enabled: true
  provider: ollama
  base_url: http://192.168.1.100:11434
  model: llava

Face Recognition (Frigate+)

人脸识别（Frigate+）

yaml
face_recognition:
  enabled: true
  threshold: 0.6

cameras:
  front_door:
    detect:
      width: 1280  # Higher res improves face detection
yaml
face_recognition:
  enabled: true
  threshold: 0.6

cameras:
  front_door:
    detect:
      width: 1280  # 更高分辨率可提升人脸识别效果

License Plate Recognition

车牌识别

yaml
lpr:
  enabled: true

cameras:
  driveway:
    lpr:
      enabled: true
yaml
lpr:
  enabled: true

cameras:
  driveway:
    lpr:
      enabled: true

go2rtc Integration

go2rtc集成

yaml
go2rtc:
  streams:
    front_door:
      - rtsp://user:pass@192.168.1.100:554/stream1
      - "ffmpeg:front_door#video=copy#audio=opus"
  webrtc:
    candidates:
      - 192.168.1.50:8555
      - stun:8555
yaml
go2rtc:
  streams:
    front_door:
      - rtsp://user:pass@192.168.1.100:554/stream1
      - "ffmpeg:front_door#video=copy#audio=opus"
  webrtc:
    candidates:
      - 192.168.1.50:8555
      - stun:8555

Docker Compose Template

Docker Compose模板

yaml
services:
  frigate:
    container_name: frigate
    image: ghcr.io/blakeblackshear/frigate:stable
    restart: unless-stopped
    shm_size: "256mb"
    devices:
      - /dev/bus/usb:/dev/bus/usb  # USB Coral
      - /dev/dri/renderD128:/dev/dri/renderD128  # Intel GPU
    volumes:
      - ./config:/config
      - ./storage:/media/frigate
      - type: tmpfs
        target: /tmp/cache
        tmpfs:
          size: 1000000000
    ports:
      - "8971:8971"  # Web UI
      - "8554:8554"  # RTSP feeds
      - "8555:8555/tcp"  # WebRTC
      - "8555:8555/udp"  # WebRTC
    environment:
      FRIGATE_RTSP_USER: admin
      FRIGATE_RTSP_PASSWORD: ${RTSP_PASSWORD}
      FRIGATE_MQTT_USER: frigate
      FRIGATE_MQTT_PASSWORD: ${MQTT_PASSWORD}
yaml
services:
  frigate:
    container_name: frigate
    image: ghcr.io/blakeblackshear/frigate:stable
    restart: unless-stopped
    shm_size: "256mb"
    devices:
      - /dev/bus/usb:/dev/bus/usb  # USB Coral
      - /dev/dri/renderD128:/dev/dri/renderD128  # Intel GPU
    volumes:
      - ./config:/config
      - ./storage:/media/frigate
      - type: tmpfs
        target: /tmp/cache
        tmpfs:
          size: 1000000000
    ports:
      - "8971:8971"  # Web UI
      - "8554:8554"  # RTSP流
      - "8555:8555/tcp"  # WebRTC
      - "8555:8555/udp"  # WebRTC
    environment:
      FRIGATE_RTSP_USER: admin
      FRIGATE_RTSP_PASSWORD: ${RTSP_PASSWORD}
      FRIGATE_MQTT_USER: frigate
      FRIGATE_MQTT_PASSWORD: ${MQTT_PASSWORD}

Bundled Resources

附带资源

Templates

模板

Located in 
templates/
:
  • docker-compose.yml
     - Production-ready compose file
  • config-minimal.yml
     - Minimal starter config
  • config-full.yml
     - Full-featured config template
位于
templates/
目录下：
  • docker-compose.yml
     - 生产环境可用的Compose文件
  • config-minimal.yml
     - 最简入门配置
  • config-full.yml
     - 全功能配置模板

References

参考文档

Located in 
references/
:
  • detector-comparison.md
     - Detector performance comparison
  • ffmpeg-presets.md
     - All available FFmpeg presets
  • mqtt-topics.md
     - MQTT topic reference
位于
references/
目录下：
  • detector-comparison.md
     - 检测器性能对比
  • ffmpeg-presets.md
     - 所有可用的FFmpeg预设
  • mqtt-topics.md
     - MQTT主题参考

Scripts

脚本

Located in 
scripts/
:
  • validate-config.sh
    - Validate config syntax before applying
位于
scripts/
目录下：
  • validate-config.sh
    - 在应用前验证配置语法

Dependencies

依赖项

Required

必需依赖

PackageVersionPurpose
Docker20.10+Container runtime
docker-compose2.0+Service orchestration
软件包版本用途
Docker20.10+容器运行时
docker-compose2.0+服务编排

Optional

可选依赖

PackageVersionPurpose
NVIDIA Container ToolkitLatestNVIDIA GPU support
Coral Edge TPU runtimeLatestCoral TPU support
软件包版本用途
NVIDIA容器工具包最新版NVIDIA GPU支持
Coral Edge TPU运行时最新版Coral TPU支持

Official Documentation

官方文档

Troubleshooting

故障排查

Camera Shows Offline

摄像头显示离线

Symptoms: Camera fps shows 0, web UI shows offline status
Solution:
bash
undefined
症状： 摄像头帧率显示为0，Web UI显示离线状态
解决方案：
bash
undefined

Test RTSP URL directly

直接测试RTSP URL

ffprobe -rtsp_transport tcp "rtsp://user:pass@ip:554/stream"
ffprobe -rtsp_transport tcp "rtsp://user:pass@ip:554/stream"

Check Docker logs

查看Docker日志

docker logs frigate 2>&1 | grep -i "camera_name"
undefined
docker logs frigate 2>&1 | grep -i "camera_name"
undefined

High CPU Usage

CPU占用过高

Symptoms: CPU consistently above 80%, system becomes unresponsive
Solution:
  1. Enable hardware acceleration (see presets above)
  2. Reduce detect fps from 10 to 5
  3. Lower detect resolution to 720p or below
  4. Add Coral TPU for object detection
症状： CPU持续高于80%，系统无响应
解决方案：
  1. 启用硬件加速（参考上方预设）
  2. 将检测帧率从10降低到5
  3. 将检测分辨率降至720p或更低
  4. 添加Coral TPU进行目标检测

No Objects Detected

无目标被检测到

Symptoms: Motion detected but no object events created
Solution:
  1. Verify detector is configured and running: check 
    /api/stats
  2. Check object filters aren't too restrictive (min_area, threshold)
  3. Ensure detect stream resolution is correct
  4. Verify objects list includes desired types
症状： 检测到运动但未生成目标事件
解决方案：
  1. 验证检测器已配置并运行：查看
    /api/stats
  2. 检查目标过滤器是否过于严格（min_area、threshold）
  3. 确保检测流分辨率正确
  4. 验证目标列表包含所需类型

Recording Not Working

录像功能不工作

Symptoms: Events show but no recordings available
Solution:
yaml
undefined
症状： 显示事件但无可用录像
解决方案：
yaml
undefined

Ensure record role is assigned

确保已分配record角色

cameras: cam1: ffmpeg: inputs: - path: rtsp://camera/stream roles: - record # Must be explicitly set record: enabled: true # Must be true
undefined
cameras: cam1: ffmpeg: inputs: - path: rtsp://camera/stream roles: - record # 必须显式设置 record: enabled: true # 必须设置为true
undefined

Setup Checklist

部署检查清单

Before deploying Frigate, verify:
  • Docker and docker-compose installed
  • RTSP URLs tested in VLC (note actual resolution)
  • Camera credentials ready for environment variables
  • Storage volume has adequate space (100GB+ recommended)
  • Shared memory size configured (64MB per camera minimum)
  • Hardware acceleration device mapped (if applicable)
  • Coral TPU device mapped (if using)
  • MQTT broker accessible (if integrating with Home Assistant)
  • Port 8971 available for web UI
  • Firewall allows required ports (8554, 8555 for streaming)
在部署Frigate前，请确认：
  • 已安装Docker和docker-compose
  • 已在VLC中测试RTSP URL（记录实际分辨率）
  • 已准备好摄像头凭据用于环境变量
  • 存储卷有足够空间（推荐100GB以上）
  • 已配置共享内存大小（每台摄像头至少64MB）
  • 已映射硬件加速设备（如适用）
  • 已映射Coral TPU设备（如使用）
  • MQTT代理可访问（如与Home Assistant集成）
  • 端口8971可用于Web UI
  • 防火墙允许所需端口（8554、8555用于流传输）