Count and detect objects from IP cameras and send results to a PLC over Modbus
RTSP streaming with real-time object detection & counting.
$200 · Free trial · 3 cameras
AI Camera HTTP Server
AI Video Analytics, RTSP-to-HTTP Streaming & Modbus Integration
1. Product Overview
AI Camera HTTP Server is an industrial camera integration and AI video analytics solution that connects RTSP-based IP cameras with SCADA systems, HMI applications, PLCs, DCS systems, web applications and other industrial software.
The software performs two functions simultaneously. First, it converts RTSP camera streams into standard HTTP video streams that can be displayed directly in a web browser or inside a SCADA application's web browser component. Second, it analyzes the live video using an AI object detection engine and makes the detected object counts available to external systems.
Detection results can be accessed through a REST API using JSON or directly from PLC and DCS systems through the built-in Modbus TCP Server .
2. How It Works
Each configured camera provides an RTSP stream to the server. The server re-serves the video as an HTTP MJPEG stream while the AI engine analyzes the video and counts detected objects by class.
3. Main Capabilities
4. AI Object Detection
The integrated AI engine analyzes each active camera stream in real time and identifies objects within the camera scene.
The detection model recognizes 80 object classes based on the COCO dataset.
The complete detection model contains 80 supported object classes.
5. AI Data for SCADA and Automation
The major advantage of AI Camera HTTP Server is that object detection results are not limited to the camera display. The detected object counts can become part of the industrial automation system.
For example, a camera monitoring a production area can detect and count objects such as people, vehicles or other supported classes. The current counts are continuously exposed through Modbus TCP registers.
6. Modbus TCP Integration
AI Camera HTTP Server includes a built-in Modbus TCP Server running on port 502 .
PLCs, DCS systems and other Modbus TCP clients can read the AI detection results directly as holding registers without requiring additional middleware or custom scripts.
Camera Register Allocation
Each camera receives a block of 100 holding registers. The register positions correspond to the object-class indexes used by the detection model.
7. Object Register Mapping
Each object class has a fixed index within the camera's 100-register block.
The complete AI model contains 80 object classes. The register index corresponds to the object-class index defined by the application.
8. REST API Integration
Detection results are also available through an HTTP REST API. Applications can request the current object counts using an HTTP GET request.
Replace <camera-id> with the numeric camera ID displayed in the application.
The API returns a JSON object containing the detected object types and their current counts.
Example format shown for illustration. Actual returned JSON depends on the objects currently detected by the camera.
9. SCADA Integration
AI Camera HTTP Server allows live video and AI data to be integrated into the same SCADA/HMI environment.
The video can be displayed using the SCADA application's web browser control, while object counts can be acquired through Modbus TCP or the REST API.
This allows operators to view the camera and use AI-generated information as part of the same operational interface.
10. Example Industrial Applications
People Counting
Monitor the number of people visible in a defined camera scene and make the count available to SCADA or PLC systems.
Vehicle Monitoring
Detect and count supported vehicle classes such as cars, trucks, buses and motorcycles.
Industrial Monitoring
Add AI-generated visual information to existing SCADA and industrial monitoring systems.
Smart Dashboards
Combine live camera video with machine data, alarms, KPIs and AI detection results.
11. Hardware and AI Processing
The AI detection engine uses DirectML and can operate with compatible DirectX 12 GPUs from NVIDIA, AMD or Intel.
CUDA or NVIDIA-specific packages such as CUDA/cuDNN are not required. When a compatible GPU is unavailable, the detection engine can fall back to CPU processing.
12. Multiple Camera Architecture
13. Deployment Options
AI Camera HTTP Server can be deployed on the same server as the SCADA application or on a dedicated server.
Same Server
The AI Camera HTTP Server and SCADA/HMI application operate on the same Windows computer.
This can be suitable for smaller installations where sufficient processing resources are available.
Dedicated AI Camera Server
The AI Camera HTTP Server operates on a dedicated Windows computer while the SCADA/HMI system remains on its own server.
This architecture can separate AI/video processing workloads from the main SCADA server.
14. Camera Setup
Step 1 — Add RTSP Camera
Enter the RTSP URL provided by the camera.
The exact RTSP path depends on the camera manufacturer and model.
Step 2 — Configure HTTP Port
Assign a unique HTTP port to the camera. Ports from 8080 upward are suitable according to the product guide.
Step 3 — Verify the Camera
Click Verify Stream to confirm that the RTSP stream is reachable from the server.
Step 4 — Start HTTP Stream
Select the camera and click Start HTTP . Multiple cameras can also be started using Start All HTTP .
Step 5 — Open the Video
15. SCADA + AI Camera Example
This type of interface allows the operator to see the actual camera scene together with AI-generated information and conventional SCADA process data.
16. Evaluation and Licensing
The unactivated version supports up to three cameras . A license tied to the server's Machine ID is required to add additional cameras.
To request activation, open the About window, copy the Machine ID and send it to:
The Machine ID is hardware-bound. If Windows is reinstalled or the server hardware is significantly changed, contact AutoNexIOT regarding activation.
17. System Tray Operation
Closing the main application window does not shut down the server. The application minimizes to the Windows system tray and continues operating.
Right-click the tray icon to restore the application or completely exit the server.
18. Important Network Considerations
The HTTP output ports must be accessible from the systems that need to display the video. The Modbus TCP server uses port 502 .
Windows Firewall rules should be configured to allow the required inbound ports.
Each camera must use a unique HTTP output port.
19. Technical Summary
AI Camera HTTP Server combines IP camera streaming, AI object detection, REST API integration and Modbus TCP into a single industrial software platform.
Connect your cameras. Detect objects. Count them. Send the results directly to SCADA, PLC and DCS systems.