All neural networks
Multiple neural networks running on a single device
Multiple neural networks running on a single device
Supports All Neural Networks
Up to 8 Full HD IP Cameras
Dual Event Panels
Current List of Neural Network Modules
(Subject to change and expansion as the system evolves)
Person
Passenger Car
Truck / Van
Bus / Trolleybus
Tram / Train
Bicycle
Animal
Face Detection
Face Identification (optional)
Lying Person Detection
Raised Hands Detection
Bent-Over Person Detection
Leg-Raising Detection
QR Code Detection
Drone Detection
Handgun Detection
Rifle Detection
Assault Rifle Detection
Mobile Phone Detection
Handbag
Bag without Handles
Plastic Bag
Suitcase
Backpack
Box Detection
Attention! Some neural network modules that require increased computing power, such as Drone Detection, Weapons Detection, and Baggage Detection, are trained only for color cameras in this configuration.
Operation with black-and-white (night mode) cameras requires both specialized training and higher processing performance. Therefore, such applications are available as dedicated solutions, including Drone Detector, Anti-Terror, Left Object Detection, and others.
If a camera switches to night-time black-and-white mode, detection of these specific objects may become inaccurate.
Reaction to Object Actions
Neural Semantics (any object movement) | Line Crossing in a Specified Direction | Zone Entry | Change in the Number of Objects | Specific Pose | Combination such as "Weapon + Hand" | Custom Logic | Face Identification from a Database (optional)
This package does not include neural networks specifically developed for narrow-use applications such as Auto Parking, License Plate Recognition, USB Cameras, and similar solutions.
According to neural network dataset size classification, the accuracy level of the included models corresponds to Small and Middle datasets.
Specialized versions use one of these neural networks trained on Large and/or Extra Large datasets. Such models operate on higher camera resolutions and require significantly greater computing resources; therefore, they are supplied separately.
Each camera can be assigned one (any) neural network block, for example, the Core Detection block (Person, Passenger Car, Truck/Van, Bus/Trolleybus, Train/Tram, Bicycle). With additional RAM installed, two or three neural network blocks can be assigned to each camera.
There are no software restrictions on the number of neural networks that can be enabled. In theory, all neural networks can be activated on all cameras. However, excessive load will reduce neural network processing speed, which may affect event detection quality.
This package includes live video monitoring and recording for up to 8 Full HD IP cameras.
There are no restrictions on camera resolution. However, higher processing loads may reduce neural network performance. The manufacturer guarantees system reliability only within the specified operating parameters.
This package includes two "Attention!" event panels (one of them dedicated to selected priority events).
This package includes AUDIO recording from IP camera microphones in AAC format.
This package includes a WEB interface with live viewing, event archive, "Attention!" panel, and system settings.
This package includes an RTSP SERVER for retransmitting camera video streams. JSON metadata streams are transmitted simultaneously with the video streams.
This package includes an EVENT LOG with object-based search and configurable time period filtering.
This package includes neural recognition quality adjustment, which automatically interprets the user's desired detection behavior through a simple left-to-right slider, without requiring configuration of complex parameters.
This package includes direct device control both locally and through the WEB interface: 4 dry-contact sensor inputs and 4 full-power relay outputs. All of them can be powered from the Videoblazer's internal power supply.
This package includes Logic Functions:
Search in the event archive is supported using the following criteria:
This package includes separate settings for each camera for activity, archiving, and third-party services:
This package includes the Neuro-Prompter – playback of pre-recorded voice messages triggered by logic-based events. Users can create their own voice messages or use the built-in announcements, including:
A vehicle is leaving the parking area.
A person has walked past the window.
People are moving through the corridor.
A person has been standing at the entrance for an extended period.
People wearing masks have entered the warehouse.
A pack of dogs has been detected near the building exit.
Two vehicles have arrived at the building.
A truck with an empty cargo compartment is passing through the checkpoint.
A truck fully loaded with sand has passed along the road.
A person has fallen within the crane operating area.
A drone has flown past the north entrance.
Armed individuals have been detected inside the warehouse.
Attention! A person has fallen. Please provide assistance!
This package includes a FREE 8-channel GOALcity software license for Windows PC with video semantic analytics.
Video streams are received from the Videoblazer at the same resolution that the Videoblazer receives from the IP cameras.
This package includes the ability to upgrade to any higher configuration with a small additional payment.
Face identification using biometric recognition, including a database of names and event reactions based on specific individuals or groups of individuals, is available as an optional feature and is not included in the standard package.
Recognition and archiving of anonymous (non-identified) faces is included in this package.