IBBVision AI Model Trainer

IBBVision AI Model Trainer is a professional desktop software that enables you to develop your own custom object detection models without writing a single line of code. Simply upload your images, label the objects, start the training process, and obtain a production-ready artificial intelligence model.

This product has been specifically developed for devices within the IBBVision ecosystem. Models you develop with IBBVision AI Model Trainer allow you to create AI Vision Analysis Modules and enhance existing modules, exclusively for use on IBBVision devices and within the IBBVision Deep Learning Surveillance software.

Offered as a licensed product, this software is a vital component of the IBBVision product family.

Why IBBVision AI Model Trainer?

Developing an AI model using traditional methods involves numerous technical steps: image collection, labeling, coding, dataset preparation, tuning training parameters, and testing. This process can take weeks and requires specialized expertise.

IBBVision AI Model Trainer consolidates all these steps into a single program. You can develop your model through an intuitive visual interface—no coding required. The models you develop integrate seamlessly into the IBBVision ecosystem and can be used directly within IBBVision Deep Learning Surveillance software.

What Can You Do?

With the program, you can develop AI Vision Analysis Modules that detect any type of object, including:

• Consumer products (cola cans, water bottles, juice boxes)

• Security elements (people, vehicles, helmets, vests)

• Industrial parts (machine components, product defects)

• Agricultural products (fruits, plants)

• Anything you need for your custom projects

You can detect multiple different objects within the same model. The modules you develop are designed for use with IBBVision Deep Learning Surveillance software.

How Does It Work?

Using the program consists of four fundamental steps:

1. Project Creation

You create a new project in the program and define the objects (classes) you want to detect—for example, "cola can" and "water bottle."

2. Image Uploading and Labeling

You upload your images into the program. For each image, you draw rectangular bounding boxes around the objects you want to detect using your mouse. You assign each box to its corresponding object class. If you have a large number of images, you can take advantage of the auto-labeling feature.

3. Model Training

Once labeling is complete, click the "Start Training" button. Configure a few settings—such as model selection and number of epochs—and start the training process. You can monitor the progress on-screen in real time. Upon completion, the most successful model is automatically saved.

4. Testing and Deployment

Test your trained model on new images, video files, or live camera feeds. If you are satisfied with the results, you can transfer the model directly to your IBBVision devices and begin using it within IBBVision Deep Learning Surveillance software.

Key Features

Zero Coding Required
Open the program and perform all operations with mouse clicks—no programming knowledge necessary.

Auto-Labeling
You don't have to label hundreds of images manually one by one. First, manually label a small set of images and train an initial model. The program then uses this model to automatically label the remaining images—you only need to correct any mislabeled ones.

Video Frame Extraction
Automatically extract image frames from video files, allowing you to quickly convert your existing video recordings into a dataset.

Active Learning
The program identifies which images your model struggles with the most and presents them to you. By correcting these images and retraining, you can continuously improve your model's performance.

Model Versioning
Every training session is saved as a separate version. You can compare the performance of different versions and revert to any previous version at any time.

Multi-Format Export Support
Export your trained models in various formats, including PyTorch, ONNX, TensorRT, CoreML, and TFLite.

IBBVision Integration
Send your trained models directly to IBBVision Edge, DLS, and DLN devices. With remote update support, you can update your devices without physically visiting them. All your developed modules work seamlessly with IBBVision Deep Learning Surveillance software.

Automatic Backup
Your projects are automatically backed up. You can restore from a backup at any time.

Who Can Use It?

The program caters to everyone—from users with no AI background to expert developers. It is specifically designed for all users of the IBBVision ecosystem.

Security Professionals – Develop person, vehicle, helmet, and vest detection models.

Industrial Users – Perform product quality control, defective product detection, and production line analysis.

Agricultural Professionals – Conduct fruit detection, diseased area identification, and crop counting.

Retail Sector – Implement product detection, shelf analysis, and inventory management.

Custom Projects – Develop tailor-made object detection models for any client-specific need.

System Requirements

The program runs on Windows 10 and Windows 11 computers.

• Processor: Intel Core i5 or equivalent (recommended: i7 or Ryzen 7)

• RAM: Minimum 8 GB (recommended: 16 GB)

• Storage: Minimum 20 GB free space (recommended: 50 GB or more)

• Graphics Card: Recommended: NVIDIA GTX 10xx series or higher, 6 GB VRAM

A graphics card (GPU) is not mandatory, but it significantly accelerates training. Without a GPU, training may take 10–20 times longer.

Where Can You Use Your Models?

You can primarily use your trained models within the IBBVision ecosystem:

• IBBVision Edge devices

• IBBVision DLS and DLN systems

• IBBVision Deep Learning Surveillance software

Additionally, by exporting the model in different formats, you can use it in:

• Your own Python projects

• Other programming languages via ONNX format (C++, C#, Java)

• Mobile devices (CoreML for iOS, TFLite for Android)

• Web applications

Why Choose IBBVision AI Model Trainer?

• Exclusive to IBBVision Ecosystem: All your developed models are fully compatible with IBBVision devices and Deep Learning Surveillance software.

• Simple: All operations are performed via a visual interface—no coding required.

• Fast: Reduces model development time from weeks to days or even hours.

• Professional: Achieve high accuracy rates with advanced YOLOv8 technology.

• Flexible: Develop models that detect any object you need.

• Secure: Your data never leaves your computer—it remains entirely under your control.

• Scalable: Suitable for everything from small experiments to large enterprise projects.

IBBVision AI Model Trainer is a proprietary product licensed under IBBVision. All rights reserved.

IBBVision AI Suspicious Bag Detection

Suspicious Bag Detection is an AI-powered analysis module that analyzes security camera footage in real time and automatically detects objects such as bags, suitcases, packages, and similar items that have been left unattended, abandoned, or forgotten in an area. This module is also referred to as "Abandoned Object Detection" or "Unattended Item Detection."

This module continuously monitors and analyzes objects within the camera's field of view. It detects when an object is left by a person, that person subsequently moves away from the object, and the object remains stationary for a specified period of time. This situation is assessed as a potential security threat, and the system immediately generates an alarm.

What Situations Does It Detect?

The Suspicious Bag Detection module can detect the following situations and scenarios:

1. Abandoned Objects:

• Unattended Bags: A person leaving their bag or suitcase behind and walking away. This poses a security risk, especially in crowded areas such as airports, train stations, and shopping malls.

• Forgotten Items: Items left absentmindedly by a person and not retrieved. These may include wallets, backpacks, shopping bags, and similar objects.

• Suspicious Packages and Boxes: Packages, parcels, backpacks, briefcases, and similar objects that have no apparent owner and are not claimed by anyone. Such items are prioritized as they may pose a security threat.

• Objects Remaining Stationary for Extended Periods: Objects that do not move for a specified duration (e.g., 5–10 minutes) and have no one nearby.

2. Unattended Object Behaviors:

• Placement and Departure: A person placing an object on the ground and then moving away from it. This behavior is the most apparent indicator that an object has been abandoned.

• Increasing Distance Between Object and Owner: The distance between a person and their belonging gradually increasing. As the person moves away, the object is classified as "abandoned."

• Unattended Item: An object with no one in its vicinity for an extended period.

3. Risky Situations Encountered in Public Areas:

• Unattended Items in High-Traffic Areas: Suspicious objects left in crowded locations such as bus stops and metro entrances.

• Event and Festival Areas: Suspicious items left at mass gatherings such as concerts, festivals, and rallies.

• Near Critical Infrastructure: Objects left close to sensitive areas such as government buildings, embassies, and energy facilities.

How Does It Work?

The Suspicious Bag Detection module utilizes deep learning and video analysis technologies. The system operates according to the following stages:

1. Object Tracking and Recognition: The system detects and classifies all objects (people, bags, suitcases, packages, etc.) in the camera image. Each object is labeled with a unique identifier, and its movements are tracked.

2. Motion and Location Analysis: The system continuously analyzes whether each object is moving and how long it has remained in the same location. Objects that remain stationary are flagged as potential "suspicious" candidates.

3. Owner-Object Relationship Analysis: When an object remains stationary, the system attempts to identify the person who left it. When that person moves away from the object and the object remains stationary for a specified period, it is classified as "abandoned."

4. Timeout Check: When an object remains stationary for a specified duration (e.g., 30 seconds, 1 minute, or 5 minutes) and its owner is not nearby, the system evaluates it as "suspicious." This duration is user-configurable.

5. Alarm and Notification: When a suspicious object is detected, the system automatically generates an alarm. The moment of the event is recorded, and security personnel are notified immediately.

Where Is It Used?

Suspicious Bag Detection is used particularly in areas where security is critical and crowds are dense:

1. Transportation Hubs:

• Airports: Terminal entrances, waiting lounges, baggage claim areas. These are among the most critical areas for detecting abandoned suitcases and bags.

• Train and Metro Stations: Platforms, waiting areas, and areas in front of ticket counters.

• Bus Terminals and Stops: Areas with high passenger traffic.

• Ports and Piers: Ship terminals and passenger waiting lounges.

2. Public Spaces and Gathering Places:

• Shopping Malls: Entrances, corridors, food courts, and seating areas.

• Museums and Exhibition Halls: Exhibition areas and entry/exit points.

• Concert and Festival Areas: Zones with high participant traffic.

• Sports Stadiums: Stand entrances, corridors, and waiting areas.

• Parks and Squares: Detection of abandoned objects in open areas.

3. Corporate and Commercial Areas:

• Corporate Buildings: Entrances, lobbies, and visitor waiting areas.

• Banks and Financial Institutions: Customer waiting areas and ATM zones.

• Hotels and Accommodation Facilities: Reception areas, lounges, and foyers.

4. Educational and Healthcare Institutions:

• Schools and Campuses: Entrances, corridors, cafeterias, and libraries.

• Hospitals: Emergency room waiting areas, entrances, and corridors.

5. Critical Infrastructure and Government Buildings:

• Public Buildings and Municipalities: Visitor entrances and waiting halls.

• Courthouses and Courts: Entrances and waiting areas.

• Energy Facilities: Entry points and perimeter fencing.

Advantages Provided

• Preventive Security: An abandoned bag or package could potentially contain explosives or pose another threat. This module detects suspicious objects at an early stage, enabling security teams to take preventive action before an incident occurs.

• Rapid Intervention: The system alerts security personnel as soon as a suspicious object is detected. Security teams can be quickly dispatched to the location of the object. In areas such as airports and stations, rapid intervention is critically important.

• Uninterrupted 24/7 Surveillance: Unlike human operators, the system operates continuously without fatigue or distraction. It remains actively monitoring during peak hours, at night, or when staffing levels are low.

• Low False Alarm Rate: Advanced AI algorithms successfully distinguish between temporarily placed items (e.g., a bag briefly set down by a person) and genuinely abandoned objects. The false alarm rate is kept to a minimum.

• Automatic Recording and Evidence Preservation: All suspicious object events are automatically recorded. The exact time, location, footage, and duration of the event are logged. These records can be used as evidence for post-incident review, investigation, and legal proceedings.

• Reduced Personnel Workload: It is practically impossible for security personnel to continuously monitor dozens of screens. The system automatically detects suspicious objects, allowing personnel to focus solely on genuine incidents. This increases workforce efficiency.

• Protection of Institutional Reputation: Early detection and prevention of security threats demonstrate that an organization provides a safe and secure environment. This positively contributes to customer, visitor, and employee satisfaction.

System Requirements

The software works in conjunction with IBBVision Deep Learning Surveillance software installed on Edge AI Vision Computer devices. It is not a standalone product.

If you have an Edge AI Vision Computer device with IBBVision Deep Learning Surveillance software already installed, no additional installation is required.

The module is licensed on an annual basis.

Frequently Asked Questions

Q: Does the Suspicious Bag Detection module also detect temporarily placed items as suspicious?

A: No. The system analyzes how long an item has remained stationary and how far its owner has moved away. Short-term placements (e.g., a person briefly setting down their bag) generally do not trigger an alarm. If an item remains stationary for a specified period and the owner moves away, the system generates an alarm. This duration is user-configurable.

Q: In which areas is the module most effective?

A: Suspicious Bag Detection delivers the most effective results in crowded and high-traffic areas such as airports, train stations, bus terminals, shopping malls, concert venues, stadiums, public buildings, and corporate facilities.

Q: How does the system determine the owner of an object?

A: The system uses motion analysis to identify the person who left the object. When a person places an object and walks away, the system tracks the relationship between that person and the object. If the person moves away and the object remains stationary for a specified period, the object is classified as "abandoned."

Q: What types of objects can be detected?

A: The system can detect a wide variety of objects, including bags, suitcases, backpacks, briefcases, packages, parcels, shopping bags, strollers, umbrellas, and more. Detection capabilities can be enhanced depending on the data on which the AI model is trained.

Q: Can the system operate in nighttime darkness?

A: Yes, with advanced image processing algorithms, it can operate under low-light conditions. However, for optimal performance, adequate lighting or the use of cameras with night vision capabilities is recommended.

Q: Through which notification mechanisms are detected suspicious objects communicated?

A: Detected suspicious objects are immediately displayed on the operator screen with visual marking. The footage of the event is recorded, and notifications are sent to the operator. Optionally, additional communication channels such as audible alarms, email, or mobile notifications can also be used. Furthermore, integration with third-party security systems is supported via webhook.

Q: Can multiple suspicious objects be detected simultaneously?

A: Yes. The system independently tracks multiple objects within the same field of view. Each suspicious object is detected separately, and individual alarms are generated.

The Suspicious Bag Detection module is a critical AI solution addressing today's increasing security needs. By providing early warning against terrorist attacks, sabotage attempts, and theft incidents, it makes a significant contribution to public safety.

Thanks to its modular structure, this module can be easily added to existing Deep Learning Surveillance systems. With its 24/7 uninterrupted operation, low false alarm rate, instant notification capabilities, and automatic recording advantages, it is one of the most valuable tools for security teams. It can be effectively deployed in all high-traffic areas, including airports, shopping malls, public transportation hubs, government buildings, and tourist facilities.

The Suspicious Bag Detection module is part of the IBBVision Deep Learning Surveillance software. The module's performance may vary depending on the hardware specifications of the Edge AI Vision Computer device, the number of cameras, and other concurrently running modules.

IBBVision AI Fight and Violence Detection

Fight and Violence Detection is an AI-powered analysis module that analyzes security camera footage in real time and automatically detects physically violent behaviors between individuals, including fights, pushing, punching, kicking, and similar actions.

This module continuously analyzes human body movements, body postures, motion speed, and physical interactions between people. It can distinguish between behaviors considered normal and those involving violence. The moment a fight or violent incident occurs, the system instantly recognizes it and sends an urgent notification to security personnel.

What Situations Does It Detect?

The Fight and Violence Detection module can detect the following types of violence and related situations:

1. Physical Violence:

• Punching: One person striking another with a fist. This is detected as a rapid and forceful arm movement.

• Kicking: One person striking another with a kick. Detected as the rapid and uncontrolled use of leg muscles.

• Pushing and Shoving: Individuals pushing, shoulder-barging, or jostling one another. These behaviors are typically seen in the early stages of a fight.

• Choking: One person grabbing or squeezing another's neck. Assessed as a serious violent act posing a life-threatening risk.

• Striking with a Hard Object: A person using an object in their hand (stick, bottle, stone, chair, etc.) against another person.

2. Group Violence:

• Mass Brawls: Uncontrolled fighting situations involving more than two people.

• Gang Confrontations: Situations where multiple groups engage in mutual violence.

• Lynching Attempts: Situations where multiple individuals attack a single person.

3. Other Violent Behaviors:

• Throwing Objects: A person angrily throwing objects or hurling items to the ground.

• Breaking Objects: Kicking, punching, or throwing items to break them.

• Defensive and Fleeing Movements: A person assuming a defensive posture or attempting to flee as the victim of violence.

• Verbal Escalation to Violence: Detecting the risk of a verbal argument escalating into physical violence. Elements such as abnormally loud voices and aggressive body language are analyzed.

How Does It Work?

The Fight and Violence Detection module utilizes deep learning, motion analysis, and human body posture analysis technologies:

1. Body Movement Analysis: The system continuously tracks the main joint points of the human body (head, shoulders, elbows, hands, knees, feet). It analyzes the relative positions of these points, their movement speed, and direction.

2. Speed and Sudden Movement Detection: The most significant difference between normal movements and violent actions is the speed and abruptness of the motion. Actions such as punches and kicks occur very rapidly. The system detects sudden, uncontrolled movements that exceed normal motion speed.

3. Distance and Interaction Between Two People: How close are two people to each other? Are they in physical contact? Does this contact involve violence? The system answers these questions through analysis. The difference between normal social distancing and violent close contact is determined.

4. Body Language and Posture Analysis: The difference between aggressive body language and normal posture is detected. Behaviors such as clenching fists, pulling back to strike, and assuming defensive positions are analyzed.

5. Audio and Noise Analysis: With integrated audio analysis, violent sounds such as shouting and screaming can also be detected (optional).

6. Time and Duration Analysis: Data such as how long a violent incident lasts and how many times it is repeated are also analyzed. Prolonged violent incidents are assessed as more critical.

Where Is It Used?

Fight and Violence Detection can be used in many areas where security is a priority and there is a risk of violence:

1. Public Spaces and Public Transportation:

• Metro and Train Stations: Detection of fights and violent incidents occurring in crowded environments.

• Bus Terminals and Stops: Detection of arguments and violent incidents among passengers.

• Squares and Parks: Monitoring violent incidents occurring in open areas.

2. Educational Institutions:

• Schools: Detection of fights, bullying, and violent incidents among students.

• Campuses: Detection of violent incidents among university students and in the vicinity of the school.

• Gymnasiums: Detection of violent incidents during or after student sports activities.

3. Healthcare Institutions:

• Hospitals: Detection of violent incidents in emergency rooms, among patient relatives, or between staff and patients.

• Psychiatric Clinics: Detection of violent behavior between patients or directed at staff.

4. Entertainment and Social Venues:

• Nightclubs and Bars: Detection of fights and violent incidents occurring under the influence of alcohol.

• Sports Stadiums: Detection of fights and violent incidents among fans.

• Concert Venues: Detection of violent incidents occurring in dense crowds.

5. Retail and Commercial Areas:

• Shopping Malls: Detection of violent incidents between customers or between customers and staff.

• Supermarkets: Detection of fights and arguments at checkout lines or in aisles.

6. Industrial Areas and Workplaces:

• Manufacturing Facilities: Detection of violent incidents between personnel or between personnel and management.

• Warehouses: Detection of violent incidents among employees.

7. Hotels and Accommodation Facilities:

• Hotels: Detection of violent incidents between guests or between guests and staff.

• Restaurants and Cafés: Detection of violent incidents among customers or staff.

Advantages Provided

• Immediate Intervention Capability: The system alerts security personnel the moment a fight or violent incident begins. Security teams are directed to the scene as quickly as possible. In violent incidents, every passing second increases the risk of injury.

• Preventive Security: The system can detect not only the moment of an incident but also aggressive behaviors exhibited prior to violence. This allows preventive measures to be taken before a violent incident occurs.

• Uninterrupted 24/7 Surveillance: Unlike human operators, the system operates continuously without fatigue or distraction. It remains active during nighttime hours or when staffing levels are low.

• Low False Alarm Rate: Advanced AI algorithms successfully distinguish between non-violent physical contact (handshakes, shoulder bumps, hugs, etc.) and actual violent incidents.

• Automatic Recording and Reporting: All violent incidents are automatically recorded. The exact time, location, footage, and duration of the event are logged. These records serve as evidence for post-incident review, legal proceedings, and reporting.

• Personnel Safety: Provides an additional layer of security, especially for healthcare workers, security personnel, teachers, and service industry employees. Enables rapid response in cases where employees are subjected to violence.

• Evidence for Insurance and Legal Proceedings: Footage of violent incidents is recorded and can be used as evidence in legal proceedings, insurance claims, and criminal investigations.

• Protection of Institutional Reputation: Rapid detection and intervention in violent incidents demonstrate that an organization provides a safe and peaceful environment. Resolving incidents before they escalate protects institutional reputation.

System Requirements

The software works in conjunction with IBBVision Deep Learning Surveillance software installed on Edge AI Vision Computer devices. It is not a standalone product.

If you have an Edge AI Vision Computer device with IBBVision Deep Learning Surveillance software already installed, no additional installation is required.

The module is licensed on an annual basis.

Frequently Asked Questions

Q: Does the Fight and Violence Detection module detect playful or joking actions as violence?

A: No. The system is trained to distinguish between playful, joking, or friendly physical contact and actual violent behavior. Accurate detection is achieved by analyzing the speed, force, frequency, and body language of the individuals involved.

Q: In which areas is the module most effective?

A: The Fight and Violence Detection module delivers the most effective results in crowded, high-risk areas such as nightclubs, schools, hospitals, stadiums, public transportation hubs, shopping malls, and industrial facilities.

Q: What happens if multiple people are fighting simultaneously?

A: The module independently tracks multiple individuals within the same field of view. Each violent incident is detected separately, and separate alarms are generated. In mass brawl situations, all parties are analyzed individually.

Q: Can it also perform audio analysis?

A: Optionally, integrated audio analysis can detect violent sounds such as shouting, screaming, and glass breaking. This feature provides an additional layer of security, particularly in areas with limited camera angles.

Q: Can the system operate in nighttime darkness?

A: Yes, with advanced image processing algorithms, it can operate under low-light conditions. However, for optimal performance, adequate lighting or the use of cameras with night vision capabilities is recommended.

Q: Through which notification mechanisms are detected violent incidents communicated?

A: Detected fights or violent incidents are immediately displayed on the operator screen with visual marking. Simultaneously, notifications are sent to the operator. Optionally, additional communication channels such as audible alarms, email, or mobile notifications can also be used. Integration with third-party systems is supported via webhook.

Q: Does this module completely eliminate the need for security personnel?

A: No. This module is a vital aid for security personnel. By removing the need for staff to continuously monitor dozens of screens, it enables them to focus solely on genuine incidents. The system alerts personnel when an event is detected, and personnel then intervene.

The Fight and Violence Detection module is a critical AI solution addressing today's increasing security needs. The rapid detection of violent incidents reduces post-incident response times, mitigates the impact of injuries, and saves lives.

Thanks to its modular structure, this module can be easily added to existing Deep Learning Surveillance systems, elevating institutional security standards. With its 24/7 uninterrupted operation, low false alarm rate, instant notification capabilities, and automatic recording advantages, it is one of the most valuable tools for security teams.

It can be effectively deployed in all environments where there is a risk of violence, including educational institutions, healthcare facilities, entertainment venues, public transportation hubs, and industrial facilities. Beyond security, the system is of great importance for preventing violent incidents and providing evidence in legal proceedings.

The Fight and Violence Detection module is part of the IBBVision Deep Learning Surveillance software. The module's performance may vary depending on the hardware specifications of the Edge AI Vision Computer device, the number of cameras, and other concurrently running modules.

Edge AI Image Processing and Video Analysis Computer for Up to 4 IP Cameras

Product Description

The EVS 04 Edge AI Vision Computer is a compact Edge Vision computer developed by IBBVision, designed for the local ingestion, processing, recording, and analysis of IP camera footage, as well as the on-site execution of IBBVision artificial intelligence applications.

The EVS 04 is built upon the edge computing approach, which enables image processing and AI applications to be performed as close as possible to the camera and data source. Without the need to continuously transmit footage to a remote server or cloud environment, all required processing is carried out directly on the device.

The system is designed for video ingestion from up to 4 IP cameras and enables existing IP camera infrastructures to be transformed into AI-powered video analysis applications. Thanks to its compatibility with RTSP- and ONVIF-compliant camera infrastructures, it allows existing systems to be integrated with IBBVision solutions without being tied to a specific camera brand.

The EVS 04 is positioned not merely as a video recorder, but as an edge computing platform on which IBBVision software and AI solutions operate.

The device is delivered with IBBVision Core software. IBBVision Core provides the system's fundamental video ingestion, live viewing, and video management infrastructure. AI-powered video analysis functions, however, are licensed independently of the standard hardware package.

Through this structure, the EVS 04 can be configured for different use cases according to the required AI functions. Analyses such as person and vehicle detection, object detection, facial recognition, license plate recognition, behavior analysis, zone intrusion, fall detection, and fire and smoke detection can be incorporated into the system through the licensing of the relevant IBBVision AI solutions.

The EVS 04 has been developed for stores, offices, restaurants, hotels, apartment and residential complex entrances, small and medium-sized businesses, warehouses, branch offices, educational institutions, and other facilities with video analysis needs of up to 4 cameras.

Edge AI Architecture

In traditional video analysis architectures, camera footage is transmitted to a central server or cloud environment, where analysis is performed. This approach can increase network traffic, particularly in structures with multiple locations and high camera counts, and creates dependency on centralized systems.

The EVS 04 reduces this dependency by performing image processing directly on-site. IP camera footage is ingested by the device, and all necessary image processing and AI inference tasks are carried out locally.

The key advantage of this architecture is the ability to perform real-time analysis without continuously sending video data to a remote system. This reduces video traffic on the network, enables low-latency analysis through local processing, and makes it possible to process sensitive video data without it leaving the facility. High levels of security and speed are achieved.

Even in the absence of an internet connection or during temporary outages, the system's core local functions can continue to operate. When licensed AI applications are configured to run on the device, analysis processes can be sustained independently of internet connectivity.

Camera and Video Infrastructure

The EVS 04 can manage video streams from up to 4 IP cameras. Camera connections are established via RTSP and ONVIF protocols.

This architecture allows existing IP camera infrastructure to be preserved. Users are not required to completely replace their camera systems solely to utilize AI video analysis; as long as existing cameras are technically compatible, they can be integrated with the EVS 04.

The number of cameras, image resolution, frame rate, and the AI models used can affect the system's total processing load. Therefore, real-time AI analysis capacity should be evaluated according to the selected camera and software configuration.

IBBVision Core

The EVS 04 is delivered with IBBVision Core software.

IBBVision Core is the foundational operating layer of the EVS 04. It provides functions such as managing camera connections, ingesting live video, viewing footage, and operating the system's core video infrastructure.

IBBVision Core is not an AI analysis package by itself. Therefore, when the product is used solely with Core software, it provides video ingestion, viewing, and video management functions dependent on product configuration; AI functions such as person, vehicle, object, or behavior analysis are not activated.

IBBVision AI Software Architecture

The AI capabilities of the EVS 04 are built upon a modular software architecture.

Deep Learning Surveillance is the software layer developed by IBBVision for running AI-powered video analysis applications. This layer provides the infrastructure required for licensed AI analysis applications to run on the EVS 04.

AI Image Analysis Solutions consist of applications and models that address specific analysis needs. Different analysis functions can be added to the system according to user requirements.

For example, one user may choose to use only person detection and zone intrusion analysis, while another may opt for a more comprehensive setup including person, vehicle, license plate, facial, or behavior analysis.

This modular structure allows the EVS 04 to be configured for various needs without being locked into a single use case.

Real-Time Analysis

When licensed AI solutions are activated, the EVS 04 can analyze camera feeds in real time to detect specified objects, events, or behaviors.

Events generated as a result of analysis can be used to trigger alarms, notifications, recordings, or other system outputs, depending on the software solution in use.

AI analysis capacity depends on variables such as the number of cameras, image resolution, frame rate, the models used, and the number of analysis functions running concurrently. Therefore, the required AI configuration for a specific application should be determined according to the system's use case.

Local Processing and Data Privacy

The EVS 04's edge architecture enables local processing of video data.

The fact that camera footage does not need to be continuously sent to an external cloud platform for analysis provides advantages, particularly in applications where security, operational privacy, and corporate data policies are critical.

Analysis results, event information, and required metadata can be stored on local systems or transmitted to authorized central systems, depending on the IBBVision software architecture in use.

This approach supports keeping control of on-site video data within the organization.

Use Cases

The EVS 04 can be used in a wide variety of environments requiring image processing and AI analysis for up to 4 cameras.

• In retail stores, it can be used for person and vehicle analysis, density measurement, zone intrusion, and security scenarios.

• In offices and business centers, it can be used to monitor entry areas, common spaces, and security zones.

• In hotels, restaurants, and cafés, it can be configured for customer density analysis, security, personnel movements, and specific behavior analysis.

• In apartment buildings and residential complexes, entry/exit areas, parking lots, common areas, and security zones can be monitored.

• In warehouses and logistics areas, personnel, vehicles, entry/exit, and designated security zones can be tracked.

• In educational institutions, healthcare facilities, and other corporate structures, custom video analysis scenarios can be implemented.

Product and Licensing Structure

The EVS 04 is a hardware product and is delivered with IBBVision Core software.

Deep Learning Surveillance and AI Image Analysis Solutions are licensed separately from the EVS 04 hardware.

Through this structure, customers can configure the system by selecting only the AI functions they need. AI capabilities are determined according to the product's hardware capacity and the selected software licenses.

The core product structure is as follows:

• EVS 04 → Edge AI Vision Computer hardware.

• IBBVision Core → Core video ingestion, viewing, and video management software.

• Deep Learning Surveillance → AI video analysis infrastructure.

• IBBVision AI Image Analysis Solutions → AI models and applications for specific analysis needs.

This architecture decouples hardware from software and allows the system to be scaled according to its intended use.

Pre-Sales Information

When the EVS 04 is purchased, the device is delivered with IBBVision Core software. AI video analysis functions are not an automatic part of the standard hardware package.

To use AI functions such as person detection, vehicle detection, object detection, facial recognition, anomaly detection, behavior analysis, or similar, the relevant IBBVision software and analysis licenses must be purchased separately.

The scope of AI functions available for an EVS 04 system is determined by the device's technical configuration and the selected software licenses.

The EVS 04 is designed to work with IP camera systems that support RTSP and ONVIF protocols. However, it is recommended to verify the compatibility of the camera model, resolution, codec, bitrate, FPS, and network infrastructure with system requirements prior to actual installation.

Internet connectivity is not strictly required for the core local viewing, recording, and licensed AI analysis functions running on the device. However, it should be noted that features such as centralized management, remote access, license validation, or data synchronization with external systems may require network connectivity.

The EVS 04 is designed to operate with the software ecosystem developed and supported by IBBVision. The installation or use of third-party software may be considered outside IBBVision's standard product support scope.

Frequently Asked Questions

Q1: What is the EVS 04?

A: The EVS 04 is an Edge AI Vision Computer platform capable of ingesting footage from up to 4 IP cameras and designed to run IBBVision AI software locally.

Q2: Does the EVS 04 come with AI analysis ready out of the box?

A: No. The EVS 04 hardware is delivered with IBBVision Core software. AI analysis functions are licensed separately through Deep Learning Surveillance and the relevant AI Image Analysis Solutions.

Q3: How many cameras does the EVS 04 support?

A: The EVS 04 supports up to 4 IP cameras.

Q4: Can existing IP cameras be used?

A: Existing RTSP- and ONVIF-compliant IP cameras can be used with the EVS 04, provided they meet the technical requirements.

Q5: Does it work without an internet connection?

A: The EVS 04's core local functions and licensed AI analyses running on the device can operate without a continuous internet connection, when properly configured. Features such as centralized management and remote access may require network connectivity.

Q6: Is the EVS 04 an NVR?

A: The EVS 04 is not positioned solely as a traditional NVR. In addition to video ingestion and management functions, the product is an Edge Vision computer platform capable of running IBBVision AI software.

Q7: Can AI models be purchased separately?

A: Yes. Different IBBVision AI analysis solutions can be licensed independently of the EVS 04 hardware. The AI functions to be used are determined according to the customer's needs.

Q8: Which AI analyses can be used with the EVS 04?

A: The available analyses depend on the licensed solution package and system configuration. Different analysis scenarios such as person, vehicle, and object detection, facial recognition, license plate recognition, zone intrusion, behavior analysis, fall detection, and fire and smoke detection can be implemented.

Q9: What is the key difference between the EVS 04 and the EVS 64?

A: The primary difference is camera capacity. The EVS 04 is designed for applications with up to 4 cameras, while the EVS 64 is developed for applications requiring higher camera counts. Both products are positioned within IBBVision's common software and AI ecosystem.

The EVS 04 Edge AI Vision Computer is a compact edge computing platform designed to bring local AI video analysis capability to existing IP camera infrastructures.

Supporting up to 4 cameras, the system combines video ingestion, video management, and the infrastructure required for local execution of IBBVision AI applications into a single device.

Thanks to its modular software and licensing structure, the EVS 04 allows only the required AI functions to be incorporated into the system. The product can thus be configured for different levels of use, ranging from basic viewing and recording needs to advanced real-time AI video analysis scenarios.

The core approach of the EVS 04 is not merely to record video, but to process it locally and transform it into meaningful information. In doing so, existing camera infrastructure can be converted into a smarter, faster, and operationally usable image processing system through the IBBVision AI ecosystem.

The EVS 04 is part of the Edge Vision System product family developed by IBBVision. The scope of hardware, software, and AI analysis features may vary depending on product configuration and the relevant licenses.

The EVS 04 is a proprietary hardware and software solution developed by IBBVision. All rights reserved.

IBBVision DLS-04
AI-Powered Video Analysis Software (4 Cameras)

IBBVision DLS-04 is an AI-powered video analysis software that, unlike traditional video management systems, does not merely record video from the field but processes it, derives meaning from it, and generates data that directly contributes to operational processes. Designed as an integral part of the IBBVision ecosystem, this software is optimized to run exclusively on IBBVision Edge AI Vision Computer hardware (EVS series).

Supporting up to four IP cameras simultaneously, DLS-04 offers an economical and efficient entry point for the video analysis needs of small and medium-sized facilities. The system is capable of interpreting live camera streams in real time. Furthermore, through one of its most powerful features—Retrospective AI—it enables AI-powered querying of historical recordings without the need for manual replay, reducing post-event analysis times to seconds and significantly decreasing operator dependency.

Important Note: IBBVision DLS-04 is not a standalone application. It requires an IBBVision Edge AI Vision Computer hardware device to operate. On its own, the software can only provide real-time viewing and event-based buffer recording for up to 4 cameras; it does not include any AI-based analysis or detection functions. To unlock advanced AI analysis capabilities (person, vehicle, object, face, license plate, zone intrusion, behavior analysis, etc.), IBBVision AI Image Analysis Module licenses must be purchased separately and activated on the system.

Deployment Model

IBBVision DLS-04 is designed to operate exclusively on IBBVision-certified Edge AI Vision Computer devices, ensuring maximum performance and compatibility for the user. This approach guarantees complete integration between software and hardware, ensuring system stability, speed, and reliability. The software is delivered pre-installed on IBBVision EVS series devices and comes ready for use.

Target Users and Application Areas

IBBVision DLS-04 offers a scalable and flexible analysis solution for small and medium-sized businesses, branch offices, apartment and residential complex management, schools, hospitals, hotels, retail stores, warehouse and logistics areas, specific sections of manufacturing facilities, and all organizations seeking to enhance their existing security infrastructure with artificial intelligence. By providing seamless integration with other modules of the IBBVision ecosystem, it delivers end-to-end security and operational analysis capabilities.

4-Channel IP Camera and Parallel Processing Architecture

The platform simultaneously processes and records video streams from up to 4 IP cameras. Each camera channel has its own independent analysis queue, and processor resources are dynamically allocated on a per-camera basis. The parallel inference infrastructure continuously analyzes motion density within the scene, optimizing frame processing frequency and preventing any bottlenecks or latency. This ensures timely alarm generation and an uninterrupted recording flow for all cameras.

DLS-04 is the entry-level model of IBBVision's DLS (Deep Learning Surveillance) product family. With its 4-camera capacity, it provides an ideal starting point for small-scale facilities. Its scalable architecture allows for seamless upgrades to higher models such as DLS-16 or DLS-32 within the same software ecosystem, should the business grow or camera requirements increase.

Real-Time AI-Powered Video Analysis (Requires Licensing)

The core software layer of DLS-04 provides video ingestion, live viewing, and event-based buffer recording functions. However, all AI-driven analysis capabilities are activated through separately purchased IBBVision AI Image Analysis Module licenses.

When these modules are licensed, independent analysis engines running on each camera are activated. The system supports enterprise-level scenarios including object detection, human and vehicle detection, multi-target tracking, zone intrusion detection, density and motion analysis, facial recognition, and license plate recognition. Additionally, the system can automatically switch between performance-focused and resource-saving operating modes based on resource usage and scene complexity.

AI Analysis on Historical Recordings (Retrospective AI)

One of the most notable and distinguishing features of IBBVision DLS-04 is its ability to apply analysis capabilities not only to live streams but also to all previously recorded video archives. Whether determining where a specific person appeared, which cameras a vehicle passed through, at what times a face was detected, or which points a particular license plate passed through—all of this can be queried in seconds using the Retrospective AI feature. The system performs these operations using feature vectors (embeddings) generated and stored during recording. This approach dramatically accelerates post-event investigations and nearly eliminates the need for security personnel to manually review hours of video footage.

Smart Search and Metadata Querying

The platform features a powerful search engine that reduces hours of video review to seconds. Users can perform smart searches by combining a wide range of criteria, including object type, color, size, movement direction, speed, and time spent in a specific area. All detected objects and events are indexed as metadata, allowing users to access this data through natural language-like queries.

Multi-Layered Security and Behavior Analysis

The comprehensive analysis infrastructure (with licensed modules) supports a wide range of security-focused use cases, including suspicious object detection, abnormal movement detection, unauthorized area access, PPE compliance monitoring (helmets, vests, masks), wrong-direction movement, vehicle-pedestrian differentiation, crowd density analysis, counting, and statistical inference. The integrated alarm engine evaluates events by time, zone, and camera, creating a prioritized and manageable alarm flow. This enables operators to focus more quickly on genuine threats.

Organization-Specific Model Development and Integration

The DLS-04 platform allows organizations to integrate an unlimited number of custom AI models tailored to their operational needs. The model lifecycle encompasses data collection, labeling, training, validation, performance measurement, and version-controlled deployment. This capability enables industry-specific risk scenarios (production defect detection, fire and smoke detection, unauthorized operation, equipment monitoring, etc.) to be securely and controllably incorporated into the system.

ROI (Region of Interest)-Based Prioritization

Within each camera's field of view, users can define specific Regions of Interest (ROI). Analysis density is increased at points marked as critical, while processing load is automatically reduced in less important or low-risk areas. This approach enables more efficient use of system resources, reduces false alarm rates, and enhances overall system performance.

Alarm and Event Management Layer

All alarms generated by the system are supported by visual marking, instant operator notifications (push notification, email, SMS), event logging, quick rewind, and direct access to the relevant keyframe. Operating within the IBBVision ecosystem, these alarms are displayed and managed in a unified interface alongside alarms from all other modules (fire, fall, license plate, etc.) on the central management console. Alarm history can be reported, prioritized, and integrated into task-based operational processes. All alarm rules can be flexibly defined by the user and updated as needed.

Reporting and Analytics Dashboard

The platform produces measurable data that supports decision-making for security and operations teams. Numerous indicators—including camera-based analysis densities, alarm distributions, model performance metrics, risk density maps, visitor counts, and after-hours activity reports—can be monitored through the advanced analytics panel. All reports can be exported in Excel, PDF, or visual formats and used in operational planning processes.

Centralized Management and Multi-Location Support

Through the IBBVision Central Management Console, DLS-04 enables monitoring and management of all cameras and analysis systems across different locations from a single point. Multiple DLS-04 devices or different models of the DLS series (DLS-16, DLS-32) can be managed through the same console. With user authorization, role-based access control, and multi-organization support, secure and organized management is maintained even in large-scale operations.

Technical Architecture and Processing Pipeline

The system architecture is built upon RTSP and ONVIF stream ingestion, timestamped frame processing, parallel inference pipeline, object and behavior analysis, embedding generation, metadata indexing, and real-time alarm engine operation. Through camera-based resource allocation, load balancing, and adaptive frame processing, stable performance is achieved during prolonged and intensive operations.

Specifications

Integration and System Compatibility

IBBVision DLS-04 offers RTSP and ONVIF compatibility independent of IP camera brand. Key integration features include:

• IBBVision Ecosystem Integration: Ability to operate on a common management console with all Edge AI modules such as EPFD-01, ESPD-01, ESPD-02, ESPD-04, EFLD-01, EFLD-02, EFLD-04.

• Camera Compatibility: Compatibility with over 200 camera brands, including all ONVIF-compliant IP cameras.

• API Services: RESTful API service outputs, Webhook-enabled alarm integrations.

• User Management: User synchronization with LDAP/Active Directory.

• Database Export: MySQL, PostgreSQL, MSSQL support.

• External System Integrations: Access control systems, fire panels, emergency systems, Building Management Systems (BMS).

Key Use Cases

• Small and Medium-Sized Businesses: Offices, stores, showrooms, restaurants, cafés.

• Residential Complexes and Apartments: Entrances, common areas, parking lots, gardens.

• Branch Offices and Bank Branches: Branch security, customer tracking, cashier areas.

• Schools and Kindergartens: School entrances, gardens, corridors, cafeterias.

• Small-Scale Hospitals and Clinics: Patient waiting areas, emergency entrances, corridors.

• Hotels and Boutique Hotels: Main entrance, reception area, parking lot, back-of-house areas.

• Retail Stores: In-store customer analysis, checkout areas, storage areas.

• Warehouses and Logistics Areas: Personnel entry, loading docks, indoor tracking.

• Manufacturing Facilities (Specific Sections): Specific production line sections, quality control points.

Why IBBVision DLS-04?

• Protects Existing Investment: Preserves your existing camera infrastructure while transforming it into an intelligent system.

• Economical Entry with 4 Cameras: Ideal starting point for small facilities, prevents purchasing unnecessary capacity.

• Scalability: Seamless upgrade path to higher models such as DLS-16 and DLS-32 as needs grow.

• IBBVision Ecosystem Compatibility: Ability to operate on a single platform with other IBBVision Edge AI modules.

• Flexibility: Organization-specific model development and integration capability.

• Historical Analysis: Not just real-time, but retrospective smart querying (Retrospective AI).

• Open Architecture: Easy integration with other enterprise systems via REST API.

• Low Total Cost of Ownership: Licensing flexibility and hardware independence.

• Centralized Management: All devices can be managed from a single point via the IBBVision Central Management Console.

• Guaranteed Performance: Superior stability and speed through jointly optimized software and hardware.

Frequently Asked Questions (FAQ)

Q1: What exactly is IBBVision DLS-04?

A: IBBVision DLS-04 is an AI-powered video analysis software that runs on IBBVision Edge AI Vision Computer (EVS series) devices, supporting up to 4 IP cameras. It offers live analysis and smart search (Retrospective AI) capabilities on historical recordings.

Q2: Can I install this software on any computer?

A: No. IBBVision DLS-04 is not a standalone software. It is designed to run exclusively on IBBVision-certified Edge AI Vision Computer (EVS-04, EVS-08, EVS-16, etc.) hardware and is delivered pre-installed on these devices.

Q3: Does DLS-04 come with AI analysis features included?

A: No. The base package of DLS-04 provides only viewing and recording functions. All AI-powered detection features—including person, vehicle, face, license plate recognition, zone intrusion, and behavior analysis—require separate purchase of IBBVision AI Image Analysis Module licenses.

Q4: How many cameras can I use with DLS-04?

A: IBBVision DLS-04 is licensed to support a maximum of 4 IP camera channels simultaneously.

Q5: Will DLS-04 work with my existing IP cameras?

A: Yes. DLS-04 supports open standard protocols such as RTSP and ONVIF. Any brand of IP camera that is compatible with these protocols and meets the technical requirements can work seamlessly with the system.

Q6: How can I search through historical recordings?

A: Using DLS-04's Retrospective AI feature, you can perform smart searches within recorded footage based on criteria such as object type, color, size, and movement direction. The system delivers results for these queries in seconds, thanks to metadata indexing.

Q7: Is a separate license required for the Retrospective AI feature?

A: The Retrospective AI feature is one of the core capabilities of IBBVision DLS-04. However, for it to function and produce meaningful results, the required AI analysis modules (object, face, license plate, etc.) must be licensed and active during recording.

Q8: Can the system operate without an internet connection?

A: Yes. DLS-04 running on EVS hardware can continue local viewing, recording, and licensed AI analyses uninterrupted without an internet connection. Additional features such as centralized management or remote access may require network connectivity.

Q9: Can I upgrade DLS-04 to support more cameras?

A: Yes. DLS-04 is the entry-level product of the IBBVision DLS family. As your needs grow, you can perform a license transition to higher models such as DLS-16 (16 cameras) or DLS-32 (32 cameras) within the same ecosystem.

Q10: What kind of technical support can I receive for DLS-04?

A: IBBVision DLS-04 is supported remotely or on-site by the IBBVision technical team, depending on the purchased license and support package. Since the software is designed to operate exclusively within IBBVision's official ecosystem, all updates and patches are also provided by IBBVision.

IBBVision DLS-04 Deep Learning Surveillance Software is a professional AI platform developed within the IBBVision ecosystem, offering an economical and efficient solution for the video analysis needs of small and medium-sized facilities. With four-camera support, it transforms existing IP camera infrastructure from a passive recording archive into an active intelligence hub.

It must be remembered that: This software is not a standalone product and requires an IBBVision Edge AI Vision Computer hardware device. The base package provides only viewing and recording functions. All AI analysis capabilities require separate purchase of IBBVision AI Image Analysis Module licenses.

In addition to live analysis capabilities, the Retrospective AI feature enables smart searching on historical recordings, reducing hours-long manual review processes for security teams to seconds. Working seamlessly with other modules of the IBBVision ecosystem, it provides the ability to manage specialized scenario solutions such as EPFD-01, ESPD-01, and EFLD-01 on a single platform. With its scalable architecture offering seamless upgrade paths to higher models such as DLS-16 or DLS-32 as needs grow, it provides future-proof investment protection.

IBBVision DLS-04 is a proprietary software solution owned by IBBVision. All rights reserved.

IBBVision DLN 24 — AI-Powered Enterprise NVR and Video Analytics Platform

IBBVision DLN 24 has been developed as an artificial intelligence–based video analytics platform that goes beyond traditional NVR systems limited to video recording. The platform actively processes and analyzes video streams received from the field, transforming visual data into meaningful operational intelligence. While simultaneously interpreting live camera feeds, the system also enables AI-assisted querying of historical recordings without the need for manual playback, significantly reducing post-incident investigation time and minimizing operator dependency.

IBBVision DLN 24 provides a scalable, stable, and enterprise-grade analytics infrastructure designed for corporate facilities, municipal areas, campus environments, industrial sites, and high-security zones.

24-Channel IP Camera Support and Parallel Processing Architecture

The platform processes and records video streams from up to 24 IP cameras concurrently. Each camera is assigned an independent analysis queue with dedicated processing resources. The parallel inference architecture dynamically evaluates scene density to optimize frame processing frequency, preventing latency under high load conditions. As a result, analytical continuity is maintained, alarms are generated in real time, and uninterrupted recording is ensured.

Real-Time AI-Based Video Analytics

Each camera operates with an independent analytics engine supporting enterprise-level scenarios such as object detection, human and vehicle recognition, multi-target tracking, individual-based tracking, intrusion detection, crowd density analysis, and motion behavior analysis. The system can automatically switch between performance-optimized and resource-efficient operating modes based on scene complexity and system resource utilization.

AI Analysis on Historical Recordings (Retrospective AI)

IBBVision DLN 24 is not limited to live stream analytics; it also enables AI-based target search on recorded video data. Queries such as where a specific individual appeared, which cameras a particular vehicle passed through, or during which time intervals a face was detected can be executed retrospectively. These operations rely on feature vectors generated during recording, accelerating post-event investigations and significantly reducing the need for manual review.

Multi-Layer Security and Behavior Analytics

The multi-layer analytics framework supports security-focused use cases including suspicious object detection, abnormal behavior recognition, unauthorized area access, personal protective equipment compliance, reverse-direction movement detection, vehicle-pedestrian classification, and density analysis. The alarm engine evaluates events by time, zone, and camera, generating a prioritized and manageable alarm workflow.

Custom Model Development and Institutional Integration

The DLN 24 platform allows institutions to integrate up to 20 custom AI models tailored to specific operational requirements. The model lifecycle includes data collection, label validation, performance evaluation, and version-controlled deployment. This structure enables secure and controlled integration of sector-specific risk scenarios into the system.

ROI (Region of Interest)-Based Prioritization

Regions of interest can be defined on a per-camera basis. Analytical intensity can be increased in critical areas while reducing processing load in non-essential zones. This approach balances system resources and helps lower false alarm rates.

Alarm and Incident Management Layer

Alarms generated by the system are supported by visual overlays, real-time operator notifications, event logging, rapid rewind functionality, and direct access to relevant video segments. Alarm histories can be reported and integrated into task-oriented operational workflows.

Reporting and Analytics Monitoring Dashboard

The platform generates measurable, decision-support data for security and operations teams. Camera-level analytics loads, alarm distributions, model performance metrics, and risk density maps can be monitored and leveraged for operational planning and optimization.

Technical Architecture and Processing Pipeline

The system architecture is built around RTSP stream ingestion, time-stamped frame processing, a parallel inference pipeline, object and behavior analysis, embedding generation, and simultaneous alarm engine execution. Camera-level resource allocation and adaptive frame processing ensure stable performance during long-term and high-intensity operations.

Integration and System Compatibility

IBBVision DLN 24 offers RTSP compatibility independent of IP camera brands. It can integrate with third-party VMS and CMS solutions and supports REST-based service outputs as well as webhook-based alarm integrations.

Key Application Areas

Corporate facilities, municipal and public spaces, manufacturing and quality control lines, commercial enterprises, energy sites, and critical infrastructure zones represent the primary application domains targeted by the DLN 24 platform.

IBBVision DLN 32 — AI-Powered Enterprise NVR and Video Analytics Platform

IBBVision DLN 32 is designed as an artificial intelligence–based video analytics platform that goes beyond conventional NVR systems limited to video recording. The platform processes, interprets, and contextualizes video streams received from the field, transforming them into actionable data for operational workflows. While simultaneously analyzing live camera feeds, the system enables AI-assisted querying of historical recordings without the need for manual playback, significantly shortening post-incident evaluation times and substantially reducing operator dependency.

IBBVision DLN 32 delivers a scalable, stable, and enterprise-grade analytics infrastructure for corporate facilities, municipal areas, production sites, campus environments, and locations with elevated security requirements.

32-Channel IP Camera Support and Parallel Processing Architecture

The platform processes and records video streams from up to 32 IP cameras concurrently. An independent analysis queue is defined for each camera, and processing resources are allocated on a per-camera basis. The parallel inference architecture dynamically evaluates scene density to optimize frame processing frequency and minimize latency. This ensures uninterrupted analytics continuity, timely alarm generation, and seamless recording workflows.

Real-Time AI-Based Video Analytics

Each camera operates with an independently running analytics engine supporting enterprise use cases such as object detection, human and vehicle recognition, multi-target tracking, individual-based tracking, intrusion detection, crowd density analysis, and motion behavior analysis. The system can automatically switch between performance-focused and resource-efficient operating modes based on scene complexity and resource utilization.

AI Analysis on Historical Recordings (Retrospective AI)

IBBVision DLN 32 is not limited to live stream analytics; it also enables AI-based target search on recorded video content. Retrospective queries can determine where a specific individual appeared, which cameras a particular vehicle passed through, or during which time intervals a face was detected. These operations are performed using feature vectors generated during recording, accelerating post-event investigations and reducing the need for manual video review.

Multi-Layer Security and Behavior Analytics

The multi-layer analytics infrastructure supports security-focused scenarios including suspicious object detection, abnormal behavior recognition, unauthorized area access, personal protective equipment monitoring, reverse-direction movement detection, vehicle–pedestrian classification, and density analysis. The alarm engine evaluates events based on time, zone, and camera, generating a prioritized and manageable alarm flow.

Custom Model Development and Institutional Integration

The DLN 32 platform allows the integration of up to 20 custom AI models based on institutional operational requirements. The model lifecycle encompasses data collection, label validation, performance evaluation, and version-controlled deployment. This enables sector-specific risk scenarios to be securely and systematically integrated into the platform.

ROI (Region of Interest)-Based Prioritization

Regions of interest can be defined for each camera. Analytical intensity is increased in critical areas while processing load is reduced in non-essential zones. This approach balances resource utilization and helps reduce false alarm rates.

Alarm and Incident Management Layer

Alarms generated by the system are supported by visual overlays, real-time operator notifications, event logging, rapid rewind functionality, and direct access to relevant moments. Alarm histories can be reported and integrated into task-oriented operational workflows.

Reporting and Analytics Monitoring Dashboard

The platform generates measurable, decision-support data for security and operations teams. Camera-level analytics loads, alarm distributions, model performance metrics, and risk density maps can be monitored and utilized in operational planning processes.

Technical Architecture and Processing Pipeline

The system architecture is built around RTSP stream ingestion, time-stamped frame processing, a parallel inference pipeline, object and behavior analysis, embedding generation, and concurrent alarm engine execution. Camera-level resource allocation and adaptive frame processing ensure stable performance during long-duration and high-intensity operations.

Integration and System Compatibility

IBBVision DLN 32 provides RTSP compatibility independent of IP camera brands. It can integrate with third-party VMS and CMS solutions and supports REST-based service outputs as well as webhook-based alarm integrations.

Key Application Areas

Corporate campuses, municipal and public areas, production and quality control sites, commercial facilities, energy zones, and critical infrastructure environments represent the primary application domains targeted by the DLN 32 platform.

IBBVision DLN 40 — AI-Powered Enterprise NVR and Video Analytics Platform

IBBVision DLN 40 has been developed as an artificial intelligence–based analytics platform that goes beyond traditional NVR solutions, evolving from a system focused solely on recording into one that processes incoming video streams, derives meaning from visual data, and delivers actionable insights to operational workflows. While simultaneously interpreting live camera feeds, the system enables AI-assisted querying without the need to revisit historical recordings, thereby shortening post-incident analysis times and significantly reducing operator dependency.

IBBVision DLN 40 provides a scalable, stable, and enterprise-grade analytics infrastructure for corporate campuses, municipal areas, production sites, campuses, and high-security environments.

40-Channel IP Camera Support and Parallel Processing Architecture

The platform processes and records video streams from up to 40 IP cameras concurrently. An independent analysis queue is defined for each camera, with processing resources allocated on a per-camera basis. The parallel inference architecture dynamically evaluates scene density to optimize frame processing frequency and minimize latency. As a result, analytical workflows remain uninterrupted, alarms are generated in a timely manner, and recording continuity is preserved.

Real-Time AI-Based Video Analytics

Each camera runs an independent analytics engine supporting enterprise use cases such as object detection, human and vehicle recognition, multi-target tracking, individual-based tracking, intrusion detection, crowd density analysis, and motion behavior analysis. The system can automatically switch between performance-focused and resource-efficient modes based on scene complexity and resource utilization.

AI Analysis on Historical Recordings (Retrospective AI)

IBBVision DLN 40 is not limited to live-stream analytics; it also enables AI-based target search across recorded video data. Retrospective queries can identify where a specific individual appeared, which cameras a particular vehicle passed through, or during which time intervals a face was detected. These queries are executed using feature vectors generated during recording, accelerating post-incident investigations and reducing the need for manual review.

Multi-Layer Security and Behavior Analytics

The multi-layer analytics framework supports security-focused scenarios including suspicious object detection, abnormal behavior recognition, unauthorized area access, personal protective equipment monitoring, reverse-direction movement detection, vehicle–pedestrian classification, and density analysis. The alarm engine evaluates events by time, zone, and camera to generate a prioritized and manageable alarm workflow.

Custom Model Development and Institutional Integration

The DLN 40 platform allows the integration of up to 20 custom AI models tailored to institutional operational requirements. The model lifecycle encompasses data collection, label validation, performance evaluation, and version-controlled deployment, enabling secure and controlled integration of sector-specific risk scenarios.

ROI (Region of Interest)-Based Prioritization

Regions of interest can be defined for each camera. Analytical intensity is increased in critical areas while processing load is reduced in non-essential zones. This approach balances resource utilization and helps reduce false alarm rates.

Alarm and Incident Management Layer

Alarms generated by the system are supported by visual overlays, real-time operator notifications, event logging, rapid rewind functionality, and direct access to relevant moments. Alarm histories can be reported and integrated into task-oriented operational workflows.

Reporting and Analytics Monitoring Dashboard

The platform produces measurable, decision-support data for security and operations teams. Camera-level analytics loads, alarm distributions, model performance metrics, and risk density maps can be monitored and utilized in operational planning processes.

Technical Architecture and Processing Pipeline

The system architecture is built around RTSP stream ingestion, time-stamped frame processing, a parallel inference pipeline, object and behavior analysis, embedding generation, and concurrent alarm engine execution. Camera-level resource allocation and adaptive frame processing ensure stable performance during long-duration and high-intensity operations.

Integration and System Compatibility

IBBVision DLN 40 provides RTSP compatibility independent of IP camera brands. It can integrate with third-party VMS and CMS solutions and supports REST-based service outputs as well as webhook-based alarm integrations.

Key Application Areas

Corporate facilities, municipal and public spaces, production and quality control sites, shopping malls, energy zones, and critical infrastructure environments represent the primary application domains targeted by the DLN 40 platform.

IBBVision DLN 48 — AI-Powered Enterprise NVR and Video Analytics Platform

IBBVision DLN 48 is designed as an AI-enabled video analytics platform that goes beyond traditional NVR systems limited to video recording. The platform processes and analyzes video streams captured from the field, transforming visual data into meaningful insights for operational workflows. While simultaneously interpreting live camera feeds, the system also enables AI-based querying of historical recordings without the need for manual playback, thereby shortening post-incident analysis times and significantly reducing operator dependency.

IBBVision DLN 48 delivers a scalable, resilient, and enterprise-grade analytics infrastructure for corporate campuses, municipal sites, production areas, large facility complexes, and high-security environments.

48-Channel IP Camera Support and Parallel Processing Architecture

The platform processes and records video streams from up to 48 IP cameras concurrently. Each camera is assigned an independent analysis queue, and processing resources are allocated on a per-camera basis. The parallel inference architecture dynamically evaluates scene density to optimize frame processing frequency and prevent latency. As a result, analytics workflows remain uninterrupted, alarms are generated in a timely manner, and recording continuity is preserved.

Real-Time AI-Based Video Analytics

Each camera operates with an independently running analytics engine supporting enterprise use cases such as object detection, human and vehicle recognition, multi-target tracking, individual-based tracking, intrusion detection, crowd density analysis, and motion behavior analysis. The system can automatically switch between performance-focused and resource-efficient operating modes based on scene complexity and current resource utilization.

AI Analysis on Historical Recordings (Retrospective AI)

IBBVision DLN 48 is not limited to live video stream analytics; it also enables AI-based target search across recorded video content. Retrospective queries can determine where a specific individual appeared, which cameras a particular vehicle passed through, or during which time intervals a face was detected. These operations are executed using feature vectors generated during recording, accelerating post-incident investigations and minimizing the need for manual review.

Multi-Layer Security and Behavior Analytics

The multi-layer analytics infrastructure supports security-focused capabilities including suspicious object detection, abnormal behavior recognition, unauthorized area access, personal protective equipment compliance, reverse-direction movement detection, vehicle–pedestrian classification, and density analysis. The alarm engine evaluates events based on time, zone, and camera, generating a prioritized and manageable alarm workflow.

Custom Model Development and Institutional Integration

The DLN 48 platform allows the integration of up to 20 custom AI models in accordance with institutional operational requirements. The model lifecycle consists of data collection, label validation, performance evaluation, and version-controlled deployment. This structure enables secure and controlled integration of sector-specific risk scenarios into the system.

ROI (Region of Interest)-Based Prioritization

Regions of interest can be defined for each camera. Analytical intensity is increased in critical areas while processing load is reduced in non-essential zones. This approach balances resource utilization and helps reduce false alarm rates.

Alarm and Incident Management Layer

Alarms generated by the system are supported by visual overlays, real-time operator notifications, event logging, rapid rewind functionality, and direct access to relevant moments. Alarm histories can be reported and integrated into task-based operational workflows.

Reporting and Analytics Monitoring Dashboard

The platform generates measurable, decision-support data for security and operations teams. Camera-level analytics loads, alarm distributions, model performance metrics, and risk density maps can be monitored and leveraged in operational planning processes.

Technical Architecture and Processing Pipeline

The system architecture is built around RTSP stream ingestion, time-stamped frame processing, a parallel inference pipeline, object and behavior analysis, embedding generation, and concurrent alarm engine execution. Camera-level resource allocation and adaptive frame processing ensure stable performance during long-duration and high-intensity operations.

Integration and System Compatibility

IBBVision DLN 48 provides RTSP compatibility independent of IP camera brands. It can integrate with third-party VMS and CMS solutions and supports REST-based service outputs as well as webhook-based alarm integrations.

Key Application Areas

Corporate facilities, municipal and public spaces, production and quality control lines, shopping malls, energy sites, and critical infrastructure zones represent the primary application domains targeted by the DLN 48 platform.

IBBVision DLN 56 — AI-Powered Enterprise NVR and Video Analytics Platform

IBBVision DLN 56 has been developed as an AI-enabled analytics platform that goes beyond traditional NVR systems limited to video recording. It processes, interprets, and contextualizes video streams captured from the field, transforming them into actionable data for operational workflows. While simultaneously analyzing live camera feeds, the system enables AI-based querying of historical recordings without the need for repeated scanning, thereby shortening post-incident analysis times and reducing operator dependency.

IBBVision DLN 56 delivers a scalable, high-capacity, and enterprise-grade, security-focused infrastructure for corporate campuses, municipal areas, manufacturing facilities, campus environments, and critical security zones.

56-Channel IP Camera Support and Parallel Processing Architecture

The platform analyzes and records video streams from up to 56 IP cameras concurrently. Each camera is assigned an independent processing queue, with computing resources allocated on a per-camera basis. The parallel inference engine dynamically evaluates scene density to optimize frame processing rates and prevent latency. As a result, analytics continuity is maintained and alarm generation remains uninterrupted.

Real-Time AI-Based Video Analytics

Each camera operates with an independently running analytics engine that delivers advanced enterprise capabilities such as object detection, human and vehicle recognition, multi-target tracking, individual-based tracking, critical area intrusion detection, crowd density analysis, and motion behavior analysis. The system can automatically switch between performance-optimized and resource-efficient operating modes based on resource availability and scene complexity.

AI Analysis on Historical Recordings (Retrospective AI)

IBBVision DLN 56 is not limited to live-stream analytics; it also enables AI-assisted target search across recorded video content. Retrospective queries can identify where a specific individual appeared, which cameras a particular vehicle passed through, or during which time intervals a face was detected. These operations are executed using feature vectors generated during recording, accelerating post-incident review and minimizing the need for manual inspection.

Multi-Layer Security and Behavior Analytics

The multi-layer analytics framework supports critical security use cases including suspicious object detection, abnormal behavior recognition, unauthorized area access, personal protective equipment compliance monitoring, reverse-direction movement detection, vehicle–pedestrian classification, and density analysis. The alarm engine evaluates events by time, zone, and camera to generate a prioritized alarm workflow.

Custom Model Development and Institutional Integration

The DLN 56 platform allows the integration of up to 20 custom AI models based on institutional field requirements. The model lifecycle encompasses data collection, label validation, performance evaluation, and version-controlled deployment, enabling secure integration of sector-specific risk scenarios.

ROI (Region of Interest)-Based Prioritization

Regions of interest can be defined for each camera. Analytical intensity is increased in critical focal areas while processing load is reduced in non-essential regions. This approach reduces false alarm probability and ensures more balanced resource utilization.

Alarm and Incident Management Layer

Alarms generated by the system are supported by visual overlays, real-time operator notifications, event logging, rapid rewind functionality, and direct access to relevant moments. Alarm histories can be reported and integrated into task-based management workflows.

Reporting and Analytics Monitoring Dashboard

The platform produces measurable data for security and operations teams. Camera-level analytics loads, alarm distributions, model performance metrics, and risk density maps can be monitored and used within decision-support processes.

Technical Architecture and Processing Pipeline

The system architecture is built around RTSP stream ingestion, time-stamped frame processing, a parallel inference pipeline, object and behavior analysis, embedding generation, and concurrent alarm engine execution. Camera-level resource allocation and adaptive frame processing ensure stable performance during long-term operations.

Integration and System Compatibility

IBBVision DLN 56 is RTSP-compatible with all IP camera brands. It can integrate with third-party VMS and CMS solutions and supports REST-based services as well as webhook-based alarm outputs.

Key Application Areas

Corporate campuses, municipal and public areas, production lines, shopping malls, energy sites, and critical infrastructure zones represent the primary application scenarios targeted by the DLN 56 platform.

IBBVision DLN 64 — AI-Powered Enterprise NVR and Video Analytics Platform

IBBVision DLN 64 is designed as an AI-based video analytics platform that goes beyond conventional NVR systems, transforming from a device that merely records video into a system that processes visual data captured from the field, analyzes it, and converts it into operational value. While analyzing live camera streams in real time, the system also examines historical video recordings using AI-based scanning techniques without the need for manual playback, accelerating post-incident analysis processes and reducing human dependency.

IBBVision DLN 64 provides a high-capacity, scalable, and security-focused analytics infrastructure for corporate facilities, public spaces, production sites, large-scale campuses, and critical infrastructure environments.

64-Channel IP Camera Support and Parallel Processing Architecture

The system processes and stores streams from up to 64 IP cameras simultaneously. Each camera is assigned an independent analysis queue, and CPU/GPU resources are allocated on a per-camera basis. Thanks to the parallel AI inference infrastructure, adaptive frame processing is applied in high-density scenes, minimizing latency. As a result, analytics workflows remain uninterrupted, alarm generation is not delayed, and recording continuity is preserved.

Real-Time AI-Based Video Analytics

Each camera runs an independent analytics engine supporting capabilities such as object detection and classification; human, vehicle, and face analytics; multi-target tracking; identity-based person tracking; zone intrusion and line-crossing detection; crowding and density analysis; and occupational safety–oriented violation detection.

The analytics engine can dynamically switch between low-latency, accuracy-focused, or resource-efficient operating modes based on scene density and the current state of system resources.

AI Analysis on Historical Recordings (Retrospective AI)

IBBVision DLN 64 is not limited to live video analytics; it also performs AI-based scanning on historical video recordings. Users can retrospectively query where and how many times a specific person appeared, which cameras a particular vehicle passed through, or where a face was detected within specific time intervals.

Instead of reprocessing archived footage from scratch, the system utilizes feature vectors generated during recording. This vector-based search approach accelerates post-incident investigations, enables direct access to critical moments, and significantly reduces the need for manual video review.

Multi-Layer Security and Behavior Analytics

Within its core model package, the system supports a wide range of object and alarm classes under security analytics, occupational health and safety, and area utilization analytics. These include suspicious object and abandoned item detection; recognition of abnormal behaviors such as running or fighting; unauthorized area access; personal protective equipment detection; risky movement recognition; vehicle–pedestrian classification; reverse-direction movement detection; and crowd/queue analysis.

The alarm generation layer evaluates events by camera, time, and zone, prioritizes alarms based on severity scores, and can trigger automated recording workflows when required.

Custom Model Development and Institutional Integration

IBBVision DLN 64 allows the integration of up to 20 custom AI models tailored to the specific risks and operational needs of different institutions and sectors. Institution-specific scenarios—such as surface deformations in municipal environments, incorrect part placement on production lines, hazardous area approaches in energy sites, or suspicious package types in shopping malls—can be incorporated into the system.

The model lifecycle includes data sampling, label validation, edge-oriented model compilation, and continuous monitoring of performance metrics. Models are integrated into the platform using version-controlled deployment.

ROI (Region of Interest)–Based Prioritization

By defining regions of interest for each camera, analysis frequency can be increased in critical areas while computational load is reduced in non-essential regions. This approach balances resource consumption, lowers false alarm rates, and enhances the effectiveness of security coverage in critical zones.

Alarm and Incident Management Layer

Alarms generated by the system are supported by visual overlays, real-time on-screen operator notifications, audible alerts, automatic event recording, and rapid access to incident moments (rewind and jump-to-event functionality). Alarm flows are prioritized by severity score, tasks can be assigned among operators, and alarm histories can be reported.

Reporting and Analytics Monitoring Dashboard

IBBVision DLN 64 does more than store recordings; it generates meaningful data for security and operations teams. Event and alarm distributions, detection accuracy rates, camera-level analytics intensity, model performance metrics, and scene-based risk maps can be monitored through the system.

These insights are used as decision-support mechanisms for preventive security planning, camera placement optimization, and workforce planning processes.

Technical Architecture and Processing Pipeline

The system architecture consists of RTSP stream ingestion, transfer to time-stamped frame queues, execution of a parallel AI inference pipeline, object and behavior analysis, generation and storage of embedding data alongside metadata, and concurrent operation of the alarm engine.

The infrastructure supports a multi-process inference architecture, camera-level CPU/GPU resource allocation, parallel queue processing, runtime model selection management, and latency-focused adaptive frame processing.

Integration and Compatibility

The system provides full RTSP compatibility with IP cameras from all manufacturers. It can integrate with third-party VMS and CMS solutions and supports REST-based service outputs as well as webhook-based alarm integrations.

Application Areas

Corporate campuses and large-scale facilities, municipal and public spaces, multi-entrance buildings and shopping malls, production and quality control lines, energy sites, and critical infrastructure environments are among the primary application areas targeted by IBBVision DLN 64.

Real-Time Intrusion Detection • 4 IP Camera Support • AI-Powered Video Analytics

IBBVision Region Intrusion Detection is a professional, AI-powered video analytics solution designed to detect unauthorized access into restricted or sensitive areas in real time using up to four IP cameras. This system enhances physical security by instantly identifying defined area breaches, notifying relevant personnel, and automatically recording the incident.

Key Features

• Real-Time Intrusion Detection: Instantly detects unauthorized entries into predefined security zones.

• AI-Powered Object Recognition: Goes beyond traditional motion detection by using deep learning to identify people and objects.

• Easy Deployment: Can be set up and activated within minutes thanks to its simplified interface and fast configuration process.

• RTSP Camera Compatibility: Fully supports all standard IP cameras using the RTSP protocol.

• Supports 4 RTSP Cameras: Simultaneous real-time analysis of four live camera feeds.

• Alarm and Recording Functionality: Generates visual/audio alerts and captures footage at the moment of intrusion.

• Lightweight System Requirements: Delivers high-performance analytics on low-spec hardware.

• Real-Time Video Processing: Enables simultaneous, continuous monitoring across all connected video streams.

Application Areas

• Factories and warehouse zones

• Military facilities and high-security perimeters

• Campuses and parking areas

• Construction sites and temporary security zones

Package Content

The IBBVision Region Intrusion Detection Software is delivered as a ready-to-use system supporting up to four IP cameras. Software installation is included in the package. However, hardware services such as camera mounting and PoE switch setup are not included.

By transforming IP cameras from passive recorders into active real-time surveillance and threat detection tools, IBBVision provides a smart, economical, and efficient solution for institutions seeking to shorten response time and improve intrusion prevention.

Real-Time Intrusion Monitoring • 3 IP Camera Support • AI-Powered Video Analytics

IBBVision Region Intrusion Detection is a professional AI-powered video analytics solution capable of detecting unauthorized access into restricted or sensitive areas in real time via three IP cameras. This system enhances physical security by immediately identifying area breaches, alerting relevant personnel, and recording each incident.

Key Features

• Real-Time Region Intrusion Detection: Instantly detects unauthorized entry into predefined security zones.

• AI-Based Recognition: Performs smart analysis based on human and object detection, going beyond traditional motion sensing.

• Easy Setup: Rapid configuration and a user-friendly interface allow deployment within minutes.

• RTSP Camera Compatibility: Fully compatible with all standard IP cameras supporting the RTSP protocol.

• Supports 3 RTSP Cameras: Processes three simultaneous live video streams for real-time monitoring.

• Alarm and Recording Functionality: Automatically triggers visual/audio alerts and captures incident footage.

• Lightweight System Requirements: Delivers high-performance analytics on low-cost hardware.

• Real-Time Video Analysis: Continuously monitors and processes live feeds from all connected cameras.

Application Areas

• Factory and warehouse zones

• Military and high-security facilities

• Campuses and parking lots

• Construction sites and temporary security areas

Package Content

The IBBVision Region Intrusion Detection Software comes bundled with support for 3 IP cameras. The software is installed as part of the package and delivered as a ready-to-use system. Hardware services such as camera mounting and PoE switch setup are not included.

By transforming IP cameras from passive recorders into active surveillance and threat prevention systems, IBBVision offers an affordable, intelligent, and effective solution for organizations aiming to detect intrusions quickly and reduce response time.

Real-Time Tracking • AI-Powered Analysis • Easy Deployment

IBBVision Region Intrusion Detection Software is a professional, AI-powered video analytics solution capable of detecting unauthorized access to restricted or sensitive areas in real-time via two IP cameras. Designed to enhance physical security, the system instantly detects predefined area intrusions, alerts relevant personnel, and records the event.

Features

• Real-Time Region Intrusion Detection: Instantly identifies unauthorized entries within predefined boundaries.

• AI-Based Recognition: Performs intelligent human and object-based analysis, going beyond conventional motion detection.

• Easy Deployment: With a simple interface and rapid configuration, the system is operational within minutes.

• RTSP Camera Compatibility: Fully compatible with all standard IP cameras that support the RTSP protocol.

• RTSP Camera Capacity: Supports up to 2 simultaneous live camera connections.

• Alarm & Recording Functionality: Provides audible/visual alerts and automatically captures visual evidence during an intrusion.

• Lightweight System Requirements: Delivers high-performance analytics with low hardware cost.

• Real-Time Video Analysis: Enables continuous monitoring and analysis of 2 live video streams.

Application Areas

• Factory and warehouse premises

• Military zones and high-security facilities

• Campuses and parking areas

• Construction sites and temporary secured zones

Package Contents

IBBVision Region Intrusion Detection Software is provided with support for 2 IP cameras. Software installation is included in the package, delivered as a ready-to-use system. Hardware services such as camera mounting or PoE switch setup are not included.

IBBVision transforms your IP cameras from simple video recorders into real-time surveillance and threat prevention tools. It is a smart, cost-effective, and efficient solution for institutions seeking rapid intrusion detection and reduced response times.