What Is RTSP and Why It Matters for AI Monitoring
RTSP (Real Time Streaming Protocol) is the industry-standard protocol for transmitting live video from IP cameras. Unlike proprietary cloud-based cameras that lock footage inside a vendor's ecosystem, RTSP cameras stream directly to any compatible software — including AI monitoring platforms that can analyse the footage in real time.
Modern AI monitoring systems work by receiving each video frame, analysing it with a vision model, and generating a structured description of what is happening. The quality of this analysis depends heavily on a stable, correctly configured RTSP stream.
Step 1: Find Your Camera's RTSP URL
Every camera manufacturer formats the RTSP URL slightly differently. Here are the most common patterns:
- Hikvision:
rtsp://<user>:<pass>@<ip>:554/Streaming/Channels/101(main stream) or.../201(sub-stream) - Dahua:
rtsp://<user>:<pass>@<ip>:554/cam/realmonitor?channel=1&subtype=0 - Reolink:
rtsp://<user>:<pass>@<ip>/h264Preview_01_main - Generic ONVIF: Use an ONVIF device manager to discover the stream URL automatically.
Step 2: Choose the Right Stream Quality
Most cameras offer a *main stream* (1080p or higher) and a *sub-stream* (typically 360p–720p). For AI analysis, the sub-stream is often preferable because:
- Lower bandwidth means more stable delivery, especially on Wi-Fi or VPN connections.
- AI models typically resize frames internally to 224×224 or 336×336 — so very high resolution adds little analytical benefit but increases CPU/GPU load.
- Lower latency between frames reduces the chance of the monitoring system falling behind.
Step 3: Enable and Test the Stream
Before connecting to any monitoring platform, verify the stream works using VLC Media Player:
- Open VLC → *Media > Open Network Stream*
- Paste your RTSP URL (including username/password if required)
- Press Play — if you see live video, the camera is correctly configured
Step 4: Network Considerations
For reliable streaming, keep cameras on the same local network as your monitoring software where possible. If you're sending streams over the internet, consider a VPN tunnel (such as Tailscale or WireGuard) rather than opening port 554 directly to the internet — exposed RTSP ports are frequently scanned and targeted by automated bots.
For monitoring sites with multiple cameras, a dedicated PoE switch with QoS settings to prioritise video traffic will prevent packet loss.
Step 5: Connect to an AI Monitoring Platform
Once your RTSP stream is stable, connecting it to an AI monitoring platform is straightforward. Platforms like AegisGates Vision AI allow you to register the RTSP URL directly from a web interface. The system will continuously sample frames, analyse activity, and generate real-time narrative descriptions — alerting you the moment anything unusual occurs.
AegisGates supports multiple cameras per account and works with any standard RTSP source, including NVR outputs and IP cameras from any brand.
Common Troubleshooting Tips
- Stream freezes or drops: Reduce stream resolution or frame rate; check network utilisation.
- Authentication errors: Some cameras require URL-encoded special characters in passwords (e.g.,
@becomes%40). - High CPU usage: Use H.264 instead of H.265 — many software decoders handle H.264 more efficiently.
- Colour issues: Make sure the camera's colour mode is not set to "Day/Night" with IR cut filter stuck in the wrong position.
Setting up RTSP cameras correctly is the foundation of any effective AI monitoring deployment. Stable streams, appropriate resolution, and secure network configuration ensure your monitoring platform can deliver consistent, accurate analysis around the clock.