Dahua XVR5116HS-I3 HD-CVI DVR (16 channels, 2 Mpix, 15 fps, 1x SATA, WizSense, IVS, HDMI, VGA, H.265+)

The WizSense series are products that have additional AI electronics responsible for analyzing live video using deep learning algorithms. This feature allows the cameras to focus on the most important purpose of monitoring, namely detecting people or vehicles. This improves the monitoring system performance, as unnecessary alarms are not sent to client applications or monitoring center.

The Dahua XVR5116HS-I3 is a state-of-the-art, 16-channel HD-CVI DVR capable of recording video from cameras up to 1080p at 15 fps. It supports HD-CVI, HD-TVI, AHD, CVBS and IP standards. Video recorded by the unit can be archived on a single 16TB hard drive. In the event of a memory overflow, automatic overwriting is triggered to prevent loss of the latest data. Built-in USB ports allow you to plug in a memory stick or external drive and copy specific sections to external storage media. The USB ports can also be used to connect compatible 3G/4G modems and external Wi-Fi cards to connect to wireless networks. Image display is possible via parallel HDMI/VGA ports (the same image is displayed). The electronics are enclosed in a compact metal housing with a modern heat dissipation system and noise-reducing fans. An application for mobile devices and computers allows easy management from anywhere.

The DVR performs image analysis based on deep learning algorithms. They reflect the structure and principles of the human brain. As in the brain, learning creates connections between artificial neurons (programs) and creates a multi-layer artificial neural network. The algorithm individually analyzes model photos and recordings, looking for common features, and then creates appropriate connections. The so-called layers, responsible for detecting a specific feature, are built this way. The more layers an algorithm has, the more accurate the analysis is. Therefore, teaching with large amounts of data is crucial.
Only multi-layer networks can collect information and perform appropriate actions, in particular automatic extraction and representation of features (patterns) present in the data, teaching the network for example to recognize faces, vehicle brands, animal species or incident, e.g. a fall.

A trained algorithm is used for video analysis. The following functions can be distinguished:

The perimeter protection function is based on detecting line crossing or trespassing in the protected area. The device distinguishes between objects that are in the area and automatically filters out false alarms caused by animals, rustling leaves, bright lights, rain or snow, etc. This ensures high alarm accuracy. In addition, it can trigger an alarm when an unwanted event is detected.

It detects faces appearing in the frame and compares with those stored in the database to make identification. Face recognition can support video systems with identification of important customers, time control, access control, visitor management and more. An alarm can be triggered when an unknown or blacklisted person is detected.

Metadata is information about features extracted from a target object. There are currently four main types of metadata in security systems: human face, human body, motor vehicle, and non-motor vehicle. Face information includes gender, age, expression, presence of glasses, mask, beard, etc. Human body information includes type of clothing, color of clothing, backpacks, etc.

Compression is a very important parameter proving the device advancement. The better the compression, the lower the bandwidth requirement with the same subjective quality. H.265 (HEVC - High Efficiency Video Coding) is currently the most efficient compression method. Compared to its predecessor (H.264), it reduces the data stream and the demand for disk capacity by half. Therefore, hard drives with a halved capacity can be used, or recordings can be stored for twice as long. Its improved version H.265+ has been optimized for use in CCTV monitoring systems. They allow for further reduction of data streams.

Traditional motion detection was based on changes in the light intensity reaching a pixel (or larger clusters of pixels) of the sensor. Any such change in light triggers detection, so the system catches the movement of objects that are irrelevant to security. It is sensitive to, for example, leaves moved by wind, flying insects, birds, shadows or precipitation.
The new motion detection system SMD can analyze detection events in terms of human or vehicle detection. Thanks to that, alarm notifications are filtered and unnecessary alarms are blocked. In a system with a compatible NVR, the recordings can be filtered in terms of human or vehicle detection. SDM+ is an enhanced version of motion detection that uses deep learning algorithms. The effectiveness of event filtering is 95%.

SmartPSS (Smart Professional Surveillance System) is an easy-to-use, free software to operate and manage Dahua products on a computer running Windows and macOS operating systems. The software allows you to conveniently perform remote system configuration and then conduct live surveillance. Its intuitive and functional interface also allows you to view and download recordings, perform intelligent search, receive alarm events, two-way audio transmission, visualize the system on an e-map or record to a computer hard drive.

DMSS is an application that enables the operation and management of Dahua products on Android and iOS mobile devices. It provides a wide range of surveillance and basic configuration options - live viewing (including multiple channels simultaneously), PTZ control, digital zoom, eavesdropping, playback of archive recordings and handling of alarm notifications. The app allows connectivity via IP addresses and via P2P cloud.