The Hikvision iDS-7632NXI-M2/X is a 32-channel DeepinMind NVR (Network Video Recorder)

designed for professional surveillance applications. It's a powerful device that utilizes deep
learning algorithms to provide advanced AI functions beyond basic video recording.

Key Features and AI Capabilities

32-channel IP Camera Input: The NVR can connect up to 32 IP cameras, with a maximum
resolution of 32 MP (megapixels). This allows you to connect all 19 of your cameras.

Deep Learning Algorithms: The "DeepinMind" series name indicates that the NVR has a
dedicated AI engine. This allows it to perform intelligent analytics directly on the NVR itself (AI
by NVR), or leverage the AI capabilities of connected cameras (AI by Camera).

Facial Recognition: This is a core feature of the iDS-7632NXI-M2/X. It can perform facial
detection, capture faces, and compare them against a face picture library. The NVR can support
up to 8-channel facial recognition for video streams and up to 24-channel for face pictures.

Perimeter Protection: This function can distinguish between people and vehicles, reducing false
alarms from other movements like rain or leaves. The NVR can support up to 12-channel
perimeter protection.

Video Structuralization: This feature allows for the structured analysis of video, making it easier
to search for specific objects and events.

Storage: The NVR has two SATA interfaces, each supporting an HDD up to 16 TB, for a total of 32
TB of storage.

Configuring the NVR with 19 Cameras and AI Functions

1. Initial Setup and Camera Connection

Physical Connection: Physically connect your 19 IP cameras to your network switch.

Add Cameras to NVR: Access the NVR's web interface. Go to "Configuration" -> "System" ->
"Camera Management" and add your 19 cameras by their IP addresses. The NVR will
automatically detect and add the cameras.

2. Configuring Face Detection and Recognition

Enable AI Resource: Before configuring face recognition, you must allocate the NVR's AI
resources. In the NVR's web interface, navigate to "Configuration" -> "System" -> "System
Settings" -> "VCA Resource" or a similar menu. Select "Face Capture" as the AI resource for the
channels where you want to enable this feature. Note: Some AI functions are mutually
exclusive, so you may need to choose between face recognition and other features like
perimeter protection on a specific channel. After saving, a reboot of the NVR is required.

Create a Face Picture Library: Go to the "Face Capture" or "Intelligent Analytics" menu. Create a
new face picture library. You can name it "Employees" or something similar. The NVR can hold
up to 16 different libraries with a total of 100,000 face pictures.

Upload Employee Face Pictures:

Manual Upload: You can upload individual photos directly through the NVR's web interface.
Ensure the images are in JPEG or JPG format and are of good quality (clear, frontal view of the
face).

Batch Import: For a large number of employees, it's more efficient to use a batch import
function. You would typically need to prepare a folder with all the employee photos and a
corresponding CSV file with employee information (name, ID, etc.). Then, you can import the
entire folder via a USB drive or from a network location.

Configure Face Detection Rules: For each of the 19 cameras, navigate to the "Face Detection" or
"Face Capture" settings.

Enable: Check the "Enable" box to turn on the feature for that camera.

Draw Detection Area: Draw a detection area on the camera's live view. This focuses the NVR's
processing power on a specific part of the scene where faces are most likely to appear, such as
an entrance.

Set Arming Schedule: Define when the face detection feature should be active.

Linkage Actions: Configure what the NVR should do when a face is detected. This can include
triggering an alarm, sending an email notification, or recording a specific event.

Testing and Troubleshooting

1. Testing Facial Recognition

Live View: Go to the live view of a camera with face detection enabled. You should see a box
appear around faces as they are detected.

Search Function: Use the "Human Search" or "Face Search" function on the NVR. You can search
for faces that match a specific picture from your library (for known employees) or for all
detected faces.

Face Matching: To test the face recognition against your employee library, have an employee
walk in front of a camera. The system should capture their face and provide a notification
indicating a "Face Match" with their information. A "Face Mismatch" notification would indicate
a stranger.

2. Common Issues and Troubleshooting

Face Detection Not Working:

Check VCA Resource: Ensure the VCA resource is correctly allocated to "Face Capture" on the
relevant channel and that the NVR has been rebooted.

Check Camera Position and Lighting: Face detection works best with clear, well-lit, frontal views
of the face. Poor lighting, shadows, or faces at sharp angles can hinder detection. Ensure the
camera is positioned correctly and the scene is adequately lit.

Network Bandwidth: The NVR's incoming bandwidth is 320 Mbps. High-resolution cameras on
all 19 channels could strain this, potentially affecting AI performance. You may need to adjust
the resolution or framerate of some cameras.

Face Library Upload Issues:

File Format: Ensure all images are in the correct format (JPEG/JPG) and the size is within the
specified limit (each picture ≤ 4 MB).

CSV File Errors: If using batch import, check the format and content of your CSV file for any
errors.

Hik-Connect Offline: This is a common issue often related to network changes. Check your DNS
settings on the NVR and ensure they are correct.
This video provides a step-by-step guide on how to configure facial recognition on a Hikvision
device, which can be adapted to the iDS-7632NXI-M2/X.