ai-nvr

name	ai-nvr
description	Deploy a local AI NVR (Network Video Recorder) on Jetson Orin using NVIDIA VST and Jetson Platform Services with DeepStream pedestrian detection. Requires Jetson Orin with JetPack 6.0 and an IP camera.

AI NVR with Jetson Orin

Deploys a local AI NVR using NVIDIA VST and Jetson Platform Services. DeepStream performs pedestrian detection; results display alongside the original video stream on the VST video wall.

Hardware: Jetson Orin device (e.g. reServer Industrial J4012), IP camera, network Software: JetPack 6.0 (L4T r36.3)

---

Execution model

Run one phase at a time. After each phase:

• If output contains [STOP] → stop immediately, consult the failure decision tree
• If output ends with [OK] → tell the user "Phase N complete" and proceed

---

Phase 1 — prerequisites check (~30 s)

Confirm JetPack 6.0:

cat /etc/nv_tegra_release
# Expected: R36 (release), REVISION: 3.0

Confirm IP camera is reachable:

ping -c 2 <CAMERA_IP>

[OK] when JetPack 6.0 confirmed and camera responds. [STOP] if JetPack version is wrong or camera unreachable.

---

Phase 2 — install Jetson Platform Services (~2 min)

sudo apt update
sudo apt install -y nvidia-jetson-services

Verify services are available:

ls /opt/nvidia/jetson/services/

Expected: directories including ingress/, vst/, etc. [OK]

---

Phase 3 — configure ingress & storage (~1 min)

Create the AI NVR nginx config:

sudo tee /opt/nvidia/jetson/services/ingress/config/ai-nvr-nginx.conf > /dev/null << 'EOF'
location /emdx/ {
    rewrite ^/emdx/?(.*)$ /$1 break;
    proxy_set_header Host $host;
    proxy_set_header X-Real-IP $remote_addr;
    proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
    access_log /var/log/nginx/access.log timed_combined;
    proxy_pass http://emdx_api;
}

location /ws-emdx/ {
    rewrite ^/ws-emdx/?(.*)$ /$1 break;
    proxy_set_header Host $host;
    proxy_pass http://emdx_websocket;
    proxy_http_version 1.1;
    proxy_set_header Upgrade $http_upgrade;
    proxy_set_header Connection "upgrade";
}
EOF

(Optional) Modify VST storage path:

sudo nano /opt/nvidia/jetson/services/vst/config/vst_storage.json
# Change data_path and video_path as needed

[OK]

---

Phase 4 — start VST services (~1 min)

sudo systemctl start jetson-redis
sudo systemctl start jetson-ingress
sudo systemctl start jetson-vst

Verify containers are running:

sudo docker ps

Verify web UI is accessible at http://<JETSON_IP>:81/

[OK] when VST web UI loads.

---

Phase 5 — download & configure AI NVR (~5 min)

Download the AI NVR package from NGC: https://catalog.ngc.nvidia.com/orgs/nvidia/teams/jps/resources/reference-workflow-and-resources

Extract:

cd <DOWNLOAD_PATH>
unzip files.zip
cd files
tar -xvf ai_nvr-1.1.0.tar.gz
cd ai_nvr

Modify DeepStream config for RTSP output. Select the correct config for your module:

• Orin NX 16GB: config/deepstream/pn26/service-maker/ds-config-0_nx16.yaml
• Orin AGX / NX8 / Nano: use the corresponding file

Add RTSP sink to the DeepStream YAML (replace the existing sink section with nvrtspoutsinkbin on port 8555).

Add the WDM environment variable in the compose file:

WDM_WL_NAME_IGNORE_REGEX: ".*deepstream.*"

[OK] when config files are modified.

---

Phase 6 — launch AI NVR (~3 min)

Select the correct compose file for your Jetson module:

cd <DOWNLOAD_PATH>/files/ai_nvr

# Orin AGX:
# sudo docker compose -f compose_agx.yaml up -d --force-recreate
# Orin NX 16GB:
sudo docker compose -f compose_nx16.yaml up -d --force-recreate
# Orin NX 8GB:
# sudo docker compose -f compose_nx8.yaml up -d --force-recreate
# Orin Nano:
# sudo docker compose -f compose_nano.yaml up -d --force-recreate

Verify all containers are running:

sudo docker ps

[OK] when DeepStream and other AI NVR containers appear.

---

Phase 7 — configure cameras via web UI (~2 min)

Open browser: http://<JETSON_IP>:30080/vst/

Add cameras manually:

• Sensor Management → Add device manually → enter IP camera RTSP URL → Submit
• Add DeepStream output stream: rtsp://<JETSON_IP>:8555/ds-test (camera name MUST contain the word "deepstream")

View results: Video Wall → Select All → Start

[OK] when video wall shows camera feeds with AI detection overlays.

---

Shutdown

cd <DOWNLOAD_PATH>/files/ai_nvr
sudo docker compose -f compose_nx16.yaml down --remove-orphans
sudo systemctl stop jetson-vst
sudo systemctl stop jetson-ingress
sudo systemctl stop jetson-redis

---

Failure decision tree

Symptom	Action
apt install nvidia-jetson-services fails	Confirm JetPack 6.0 is installed. Run sudo apt update first.
VST web UI not loading on port 81	Check sudo systemctl status jetson-vst. Verify firewall allows port 81.
Docker compose fails	Verify Docker + nvidia-container-runtime installed. Check disk space.
DeepStream container exits immediately	Check sudo docker logs <container>. Verify correct compose file for your module.
No detection overlay on video wall	Confirm DeepStream config has RTSP sink on port 8555. Camera name must contain "deepstream".
Camera not visible in VST	Verify camera RTSP URL is correct and reachable from Jetson.

---

Reference files

• references/source.body.md — Original Seeed wiki with screenshots and detailed config examples