Back to Learning Matrix
M4

Edge Vision AI

Based on lightweight edge cameras and industrial multi-channel AI compute hosts, build edge vision solutions for object detection, zone intrusion alerts, and automated linkage.

Traditional surveillance relies on post-incident playback and cannot generate structured alerts when events occur; conventional motion detection suffers frequent false alarms from lighting changes and weather. Custom visual AI development has long cycles, requiring dedicated algorithm engineers and specialized rack servers, making it hard for solution providers to balance performance and hardware costs between edge endpoints and centralized servers.

Partnership Inquiry
reCamera 2002 HQ PoE 8GB Open AI Camera
Difficulty
Advanced
Duration
L1 1 day / L2 2–3 days / L3 3–5 days
Shortest Format
1 day (Taster Session · L1)
Teaching Format
3 tiers: Taster / Workshop / Bootcamp
Dual Hardware Mainlines
reCamera (Lightweight Node) + Jetson Orin NX (Heavyweight Node)
Core Protocols
RTSP / MQTT / PoE 802.3af/at / REST API
Track
Build AI Products
Who This Course Is For
Property and campus security and operations engineers · Industrial manufacturing safety management personnel · Smart commerce and foot-traffic analytics practitioners · Agriculture and outdoor monitoring system integrators · AIoT and computer vision development engineers · Faculty and students at vocational colleges and applied universities
Prerequisites
L1 requires basic networking and browser skills (connecting to Wi-Fi and accessing web UIs); L2 requires understanding IP networking, Docker, and MQTT basics, with the ability to edit YAML files; L3 requires Python and Linux command-line foundations and basic object detection concepts.

Typical Scenarios

  • Property and campus security: perimeter defense, zone-crossing alerts, nighttime abnormal loitering detection
  • Industrial manufacturing and safety: hard-hat/reflective-vest non-compliance detection, restricted-zone intrusion, conveyor belt status monitoring
  • Smart Commerce & Foot Traffic Analytics: Zone Headcount, Dwell Time Statistics, Heat Zone Distribution
  • Agriculture & Outdoor Monitoring: PoE-Powered Outdoor Point Monitoring, Specific-Target Recognition & Boundary-Crossing Alerts
  • Temporary Deployment & Mobile Workstations: Single-Point reCamera Quick Deployment, On-Device Inference Plug-and-Play

Three depths and where each one gets you

  1. L1 Demo Level 1 day

    Dual-Track Experience and Basic Configuration

    Set up one reCamera, draw a zone and detect targets

    • Understand core vision concepts such as frame rate, resolution, confidence threshold, and IoU
    • Independently complete reCamera device network configuration and video stream output
    • Master the applicable boundaries and selection criteria for lightweight-node and heavyweight-node solutions
    • Complete network configuration and basic object detection verification for one reCamera node
  2. L2 Consultant Level 3 days

    Dual-Track Scenario Automation and Multi-Channel Aggregation

    Feed multiple cameras into Frigate, cut false alarms and trigger sound-and-light alerts

    • Master using Node-RED on reCamera for local edge-event linkage
    • Master Frigate frigate.yml multi-channel configuration, zone drawing, and parameter tuning
    • Master Frigate and Home Assistant linkage configuration and cross-system event-driven logic
    • Complete one full system with 2+ RTSP channels, Frigate detection, HA linkage, and false-alarm tuning
  3. L3 Design Level 5 days

    Model Customization and Edge Deployment Optimization

    Train your own detection model and deploy it to edge hardware

    • Master the complete engineering loop of vision AI from data annotation and model training to edge deployment
    • Master the key toolchain for TensorRT and embedded model quantization conversion
    • Capable of independently designing and delivering vertical-industry vision recognition solutions
    • Deliver one custom-trained object detection model and complete on-hardware runtime verification

What learners can do afterwards

  • Configure reCamera on-device object detection and RTSP streaming
  • Orchestrate Node-RED vision events with audio-visual alert linkage
  • Build Frigate multi-channel RTSP aggregation and NVR recording management
  • Configure detection zones (Zones) and confidence threshold tuning
  • Integrate Home Assistant cross-system automation linkage
  • Train custom YOLO models and deploy with TensorRT / cvimodel quantization

Hardware and toolchain

This course centers on "open-source AI cameras + edge inference boxes", covering the full pipeline from capture and inference to alerts.

  • CAM-L reCamera 2002w/2002 Lightweight Edge Node · On-Device NPU Inference · RTSP Streaming
  • CAM-H reCamera 2002 HQ PoE 8GB PoE Primary Camera Position · Multi-Channel NVR Aggregation
  • EDGE reComputer RK3576-30 Lightweight Inference Host · 6 TOPS NPU · Node-RED Linkage
  • NVR reServer Industrial J4012 Jetson Orin NX 16GB · Frigate Multi-Channel Aggregation · TensorRT
  • NET 8+2-Port Gigabit PoE Switch 802.3af/at · PoE Power & Traffic Aggregation
Illustrated list of 6 devices
  • reComputer RK3576-30 Edge AI Box

    reComputer RK3576-30 (100052518)

    Lightweight vision inference host, 8GB RAM, 6 TOPS compute

    Edge intelligence controller running event-triggered automation and lightweight video monitoring dashboards. Equipped with 8GB RAM and 6 TOPS NPU compute, suitable for L1/L2 lightweight vision inference and Node-RED event联动. Can serve as a lightweight deployment host for Frigate and Home Assistant.

  • reCamera 2002w/2002 Open AI Camera

    reCamera 2002w/2002 (102991896/102991894)

    Single-camera vision node, on-device lightweight object detection and RTSP streaming

    Modular open-source AI camera; 2002w supports Wi-Fi/AP mode, 2002 supports 100Mbps wired Ethernet. Built-in NPU inference and Node-RED zero-code orchestration, plug-and-play, suitable for L1 single-point lightweight node experience and L2 on-device automation. Supports RTSP video stream output (port 554).

  • reCamera 2002 HQ PoE 8GB Open AI Camera

    reCamera 2002 HQ PoE 8GB (100029708)

    Multi-camera primary camera, PoE single-cable power and streaming

    Modular open-source AI camera, supports PoE switch single-cable power and video streaming. 8GB RAM configuration, suitable as the primary camera position for multi-channel NVR aggregation analysis. Connects to Frigate via RTSP, supports detection zone and confidence threshold configuration.

  • reCamera Pro AI Camera

    reCamera Pro AI Camera (100092895)

    High-Performance Open-Source AI Vision Camera, Advanced Vision Development & Higher-Compute Model Verification

    Optional expansion device, high-performance model in the reCamera series, suitable for advanced vision development and higher-compute model verification. Can handle more complex on-device inference tasks and multi-model parallel verification, serving as an advanced compute option for L2/L3 stages.

  • reServer Industrial J4012 Jetson Orin NX 16GB

    reServer Industrial J4012 (Jetson Orin NX 16GB, 114110247)

    High-performance video hub, industrial-grade Jetson edge server

    Industrial-grade Jetson Orin NX 16GB edge computing server, running Frigate NVR for multi-channel detection aggregation. Supports GPU hardware decoding and TensorRT quantization acceleration, suitable for L2 multi-channel video stream aggregation and L3 custom model inference deployment. Requires flashing JetPack image and configuring Docker --runtime nvidia; Frigate uses the stable-tensorrt-jp6 image.

  • 8+2-Port Gigabit PoE Industrial Switch

    8+2-Port Gigabit PoE Industrial Switch (802.3af/at)

    Centralized power supply and high-speed networking

    8+2-port gigabit PoE switch, compliant with 802.3af/at standards. Powers PoE cameras and aggregates LAN traffic, suitable for multi-channel reCamera HQ PoE centralized deployment and Frigate NVR aggregation scenarios.

Additionally includes CUDY AX3000 Wi-Fi 6 Gigabit Router, mini tripod × 3 (114993412), portable 13.3" 1080P display (mini HDMI and Type-C cables included), Cat6 patch cords × 4 (1m), Bull power strip (3 AC + 3 USB ports), and scenario simulation materials (hard hat and safety vest stickers, workpiece test models).

reCamera Edge Inference, Frigate Multi-Channel Aggregation, YOLO Custom Models

From plug-and-play single-point edge detection to multi-channel NVR analysis and quantized custom-model deployment — three depths of a vision solution.

  1. reCamera + Node-RED

    Edge NPU inference + no-code flow · RTSP / MQTT

    1. Device Power-On & Network Connection
    2. RTSP Video Stream Output
    3. Pre-Trained Model Object Detection
    4. Node-RED Event Linkage

  2. Frigate NVR + Home Assistant

    Multi-channel RTSP hardware decode + zone detection + MQTT alert

    1. Multi-Channel RTSP Access
    2. Zones/Masks Configuration
    3. Confidence Threshold Tuning
    4. HA Automation Linkage

  3. YOLO + TensorRT / cvimodel

    Data Annotation + Transfer Learning + Quantized Deployment

    1. Dataset Collection & Annotation
    2. YOLO Transfer Training
    3. ONNX Export
    4. TensorRT/cvimodel Quantization
    5. Edge Device Runtime Verification

Key Pivot · From Single-Node Edge Inference to Multi-Channel Centralized NVR Aggregation reCamera handles plug-and-play on-device detection and local linkage; Frigate and Jetson provide multi-stream aggregation, unified zone rules, and cross-system alerting, transitioning from single-point sensing to centralized multi-stream analytics.

Additionally requires Mosquitto MQTT Broker (structured alert events, port 1883), InfluxDB + Grafana (L3 detection statistics and time-series dashboard), and CVAT / Roboflow (L3 dataset annotation platform).

Syllabus and formats

15 teaching units in total. The delivery formats come first; the full syllabus expands below.

  • Taster Session

    No final project

    1 Day · 6–8h · L1 Foundation · Focused on reCamera single-node setup and dual-track architecture overview

  • Hands-On Course

    Includes final project

    2–3 Days · 14–20h · L1+L2 · Node-RED alert linkage + Frigate multi-stream aggregation + HA automation + false alarm tuning

  • Delivery Course

    Includes final project

    3–5 days · 24–35h · L1+L2+L3 · full coverage including custom model training and edge deployment optimization

PoE switch and reCamera HQ PoE wiring requires confirmation that switch ports support 802.3af/at standards; non-standard PoE injectors may damage camera network ports. Connecting non-PoE ports as PoE power supply is strictly prohibited.

Jetson reServer J4012 requires flashing JetPack image and configuring Docker --runtime nvidia; Frigate must use the stable-tensorrt-jp6 image, ordinary CPU images cannot enable GPU hardware decoding and TensorRT acceleration.

The taster session does not include Frigate multi-channel configuration or Node-RED hands-on. Do not promise clients that taster session students can independently complete multi-channel NVR aggregation — that is the workshop delivery standard.

Compliance red line: all detection models and scenarios in this course are limited to object/behavior/zone detection; facial identity recognition and biometric tracking are strictly prohibited. Detection boundaries must be confirmed in writing with the client before project delivery.

One of the lessons · 01

Pre-class Preparation and Environment Pre-check

Hardware bench inventory, network environment configuration, reCamera firmware pre-flashing and Node-RED/SSCMA plugin verification, Frigate container deployment and GPU passthrough configuration, teaching material distribution

—

Show the full syllabus of 15 units and the schedule for each format
15teaching modules coverage across formats
Module / OutputTaster1 dayWorkshop2–3 daysBootcamp3–5 days
01Pre-class Preparation and Environment Pre-checkHardware bench inventory, network environment configuration, reCamera firmware pre-flashing and Node-RED/SSCMA plugin verification, Frigate container deployment and GPU passthrough configuration, teaching material distribution—FullFullFull
02Vision AI Core Concepts & Dual-Track ArchitectureFrame rate FPS, resolution, confidence threshold, IoU; reCamera lightweight node vs Jetson heavyweight node technical metrics, cost, and applicable boundary comparison—FullFullFull
03reCamera Single Lightweight Node ConfigurationPower-on and network connection (2002w Wi-Fi/AP, 2002 100Mbps wired, HQ PoE switch power), web interface access, RTSP video stream verification (port 554), pre-trained model switchingreCamera Web UIFullFullFull
04Basic Intrusion Detection & Zone DrawingDraw basic detection boxes, observe target entry triggering events, observe on-device NPU inference frame rate and confidence changesreCamera / SSCMAFullFullFull
05Jetson Multi-Channel NVR Architecture DemoFrigate multi-channel RTSP access and GPU hardware decoding demo, object detection and recording management, intrusion detection → MQTT push → audio-visual alert linkage demoFrigate NVRPartialFullFull
06Node-RED Vision Event OrchestrationSSCMA model node configuration, confidence filtering and object type filtering, Dashboard real-time view and alert status dashboard constructionNode-RED / SSCMANoneFullFull
07Multi-Channel Alert LinkageMQTT message publishing drives smart bulb color change and buzzer sounding, Webhook node pushes alerts to enterprise WeChat group botNode-RED / MQTTNoneFullFull
08Frigate Multi-Camera Configurationfrigate.yml configuration structure analysis, multiple reCamera RTSP address addition, detection frame rate configuration (5–10FPS), Zones and Masks definitionFrigate / YAMLNoneFullFull
09Home Assistant Deep IntegrationFrigate HA integration plugin installation and configuration, camera entity and sensor status mapping, time-based and zone-intrusion-based automation YAML script writingHome Assistant / YAMLNoneFullFull
10False-Alarm Rate Tuning Hands-OnConfidence threshold adjustment (Min Score/Threshold), Zones coordinate optimization to exclude tree branch swaying/reflection/background clutter, pre/post-tuning false-alarm data comparison recordsFrigateNoneFullFull
11Dataset Collection & AnnotationField data collection strategy (lighting changes/multi-angle sources/positive-negative sample balance), CVAT or Roboflow object detection bounding box annotation, dataset split 7:2:1 and data augmentationCVAT / RoboflowNoneNoneFull
12YOLO Model Transfer LearningYOLO object detection principle and backbone network analysis, PyTorch training environment and pre-trained weight configuration, Loss convergence curve and mAP@0.5 metric monitoringYOLO / PyTorchNoneNoneFull
13Edge Model Conversion & DeploymentONNX format export, Jetson TensorRT Engine conversion, reCamera TPU-MLIR INT8 quantization to cvimodel, on-device inference latency/FPS/VRAM measurementTensorRT / TPU-MLIRNoneNoneFull
14Structured Data Aggregation & DashboardsObject detection statistics written to InfluxDB time-series database, Grafana alert frequency statistics/zone heatmap/compliance rate trend dashboard constructionInfluxDB / GrafanaNoneNoneFull
15Solution Review and Delivery SummaryGroup project solution rehearsal and false-alarm tuning effect defense, edge compute overhead and network bandwidth usage retrospective, deliverables and configuration file archiving—PartialFullFull

● Full◐ Partial— None●＋ Extended

Taster focuses on L1 dual-track experience and base configuration, with Jetson NVR as an observation demo without hands-on lab; workshop covers full L1+L2 Node-RED automation and Frigate multi-stream aggregation; bootcamp provides end-to-end L1+L2+L3 coverage including custom model training and edge deployment.

  • Taster Session

    • Day 1 MorningModules 01 + 02 + 03

      Pre-check → Vision AI core concepts → reCamera single-node setup

    • Day 1 AfternoonModules 04 + 05 (demo) + 15 (abbreviated)

      Zone intrusion detection → Jetson NVR architecture demo → Review & debrief

    Jetson multi-channel NVR is an observational demo; hands-on practice is not included.

  • Hands-On Course

    • Day 1Modules 01–05

      Pre-check → Vision concepts → reCamera setup → Basic detection → Jetson NVR demo

    • Day 2Modules 06–08

      Node-RED vision event flow → Multi-channel alert linkage → Frigate multi-camera setup

    • Day 3 (optional)Modules 09 + 10 + 15

      HA deep integration → False alarm tuning lab → Architecture review & delivery wrap-up

    The workshop delivers one complete system with 2+ RTSP channels, Frigate detection, HA linkage, and false-alarm tuning. Student prerequisite: understanding of IP networking, Docker, and MQTT fundamentals, ability to edit YAML configuration files.

  • Delivery Course

    • Day 1–2Modules 01–10

      Complete L1+L2 contents (reCamera setup + Node-RED linkage + Frigate multi-stream aggregation + HA integration + false alarm tuning)

    • Day 3Modules 11 + 12

      Dataset collection & annotation → YOLO model transfer learning

    • Day 4Modules 13 + 14

      Edge model conversion and deployment → Structured data aggregation and dashboard construction

    • Day 5Module 15

      Solution Review and Delivery Archiving

    The bootcamp goal is the ability to independently design and deliver vertical-industry vision recognition solutions. Student prerequisite: Python and Linux command-line fundamentals, familiarity with basic object detection principles, familiarity with L1–L2 competencies.

The taster course takes one day, needs no model training, and stops at "the camera streams, targets are detected and alerts fire". It suits trade shows, open days and first meetings with a client.

Workshop Day 3 is an optional flexible day: if students have a strong foundation, it can be compressed to 2 days (Day 2 afternoon merged with HA integration and false-alarm tuning); if more Frigate configuration tuning time is needed, use the full 3 days.

What You Take Away

The following are full-version (bootcamp) deliverables; the workshop delivers the first 4 items; the taster session delivers abbreviated versions of items 1 and 2.

  1. 01

    Vision system architecture design diagram and network topology documentation

    Includes dual-hardware-mainline topology diagram, on-site LAN IP address planning table, PoE power supply scheme, RTSP stream addresses and port allocation (554/5000/8123/1883).

  2. 02

    reCamera Node-RED linkage flow configuration files

    Includes SSCMA model node configuration, confidence and object type filtering rules, MQTT publish node, Webhook alert push flow JSON.

  3. 03

    Frigate configuration file (frigate.yml) and HA automation scripts

    Includes multi-channel RTSP camera configuration, Zones/Masks zone definitions, detection frame rate and confidence threshold, Home Assistant automation YAML rules.

  4. 04

    False-alarm rate pre/post-tuning test comparison record

    Includes pre-tuning false-alarm frequency statistics, confidence threshold and Zones adjustment records, post-tuning false-alarm rate comparison data, lighting and obstruction condition notes.

  5. 05

    Custom dataset, training configuration, and quantized model files (L3)

    Includes annotated dataset (Train/Val/Test 7:2:1), YOLO training configuration and weights, ONNX export files, TensorRT Engine (Jetson) and cvimodel (reCamera) quantized models, inference latency and FPS benchmark test report.

  6. 06

    InfluxDB Time-Series Data & Grafana Dashboard Configuration (L3)

    Includes detection statistics data table structure, Grafana alert frequency/zone heatmap/compliance rate trend dashboard configuration JSON.

Open-source projects you can reproduce

These projects use the same hardware and methods as the course, and what we learn in class is fed back into them.

  • Intrusion and loitering alerts

    One RTSP camera yields restricted-zone, line-crossing and loitering alerts, worked through in a browser. Detection runs on the edge box itself; a single Jetson or RK3588 runs the whole stack.

  • Parking slot monitor

    See which parking slots are taken and which are free in real time. It runs on a single reCamera with no extra server.

  • Retail people-flow heatmap

    See which shelves draw customers and which aisles are ignored. Works with reCamera or with IP cameras plus an AI box.

Take vision models beyond lab benchmarks to handle real lighting and false alarms

M4 focuses on edge camera deployment, edge NPU inference, multi-stream video aggregation, and on-site false alarm tuning, guiding learners through the entire delivery workflow. With hands-on lab testbeds progressing from basic to advanced, instructors can flexibly configure lightweight single-node setups or centralized multi-stream architectures based on course length.

  • Extension 01

    Adapt to your site conditions

    Units 04 and 08 enable learners to define customized detection zones, such as lab entryways, hazardous machinery perimeter, or safety corridors, calibrating zones to physical constraints.

  • Extension 02

    Connect Cameras

    Existing institutional ONVIF/RTSP surveillance cameras can be integrated into the multi-stream NVR, validating the system against real-world heterogeneous video sources.

  • Extension 03

    Integrate Teaching Experience

    Introduce real-world conditions such as sudden lighting changes, swaying branches, and glare into lab training, guiding students to tune confidence thresholds and masking zones to master false-alarm mitigation.

Scope Boundaries & Compliance

Only object/behavior/zone detection and event alerts; facial identity recognition and biometric tracking are strictly prohibited.

In Scope

  • General and Specific Object Detection (person/vehicle/hard-hat/reflective-vest/workpiece and other object categories)
  • Zone Intrusion & Boundary-Crossing Detection, Nighttime Abnormal Loitering Detection
  • Multi-Channel RTSP Video Stream Aggregation Analysis & NVR Recording Management
  • Structured Detection Events Pushed to Audio-Visual Alerts & Third-Party Systems via MQTT/Webhook
  • Zone Headcount & Dwell Time Statistics (Anonymous Aggregation, Not Linked to Personal Identity)
  • Custom Object Detection Model Training & Edge Quantized Deployment (L3)

Out of Scope

  • Facial identity recognition and biometric tracking are strictly prohibited (compliance red line); all models and scenarios in this course are limited to object/behavior/zone detection
  • Strictly prohibited for millisecond-level life-safety braking control (e.g., autonomous driving, medical diagnosis, industrial safety interlocks)
  • Does not replace statutory security monitoring systems and fire alarm systems
  • Does not promise 100% recognition rate in extreme weather (heavy rain, dense fog, strong backlight)
  • Does not include long-term on-site outsourced operations and maintenance services
  • A single reCamera is a single-camera lightweight solution supporting 1 channel of 1080P real-time inference; multi-channel concurrency requires deploying an edge host (RK3576 / J4012) and properly planning detection resolution and frame extraction rate (typically 5–10 FPS)

Put M4 on your timetable

Budgets are based on participant count and hardware requirements. Email us your expected group size and target depth to receive a detailed syllabus and kit plan within 3 business days.

Partnership Inquiry View Courses