Security teams need cameras that help them recognize a developing problem, understand what happened, and respond with confidence. Those needs become harder to meet when nighttime video loses important details, nuisance alarms consume attention, or a camera is pointed away from the action.
The next generation of IQSIGHT security cameras addresses these problems through improvements in image quality, video analytics, cybersecurity, and camera design. Together, those capabilities can give end users clearer incident evidence, fewer nuisance alerts to investigate, better visibility across a site, and more manageable system expansion.
At the Security Technology Forum 2026, Matthew Cirnigliaro, Head of Product Marketing, and Gijs van den Borne, Global Product Manager of Intelligent Fixed Cameras & Accessories, explained that connection in their presentation, “What the CPP? Inside the Next Generation of Camera Intelligence.” IQSIGHT is formerly Bosch Video Systems. Their discussion covered the technology inside the cameras, the practical problems it can address, and the design decisions that determine results.
CPP stands for Common Product Platform. It is the underlying collection of processors and related technology shared across multiple camera models. That platform determines how much work a camera can perform, including improving images, analyzing activity, protecting data, and delivering video to different users.
Compared with CPP14, they described CPP16 as providing twice the general processing power, twice the video encoding performance, and three times the AI processing performance. The additional capacity supports improvements that would otherwise compete for limited resources inside the camera.
CPP15 is optimized for cameras with multiple imagers, particularly four-sensor designs. CPP16 serves single- and dual-imager applications. They share similar processing and security foundations, with each tailored to its camera design. The platform number alone should not determine a purchasing decision.
A camera may produce an impressive daytime view and still struggle when a person moves through a dark parking area or approaches an entrance with bright sunlight behind them. In those situations, useful detail depends on how the camera handles light, movement, and image noise.
Matt emphasized that image quality supports everything that follows, including perimeter protection, traffic analysis, and reviewing safety incidents such as slips and falls. If important details disappear from the image, analytics have less reliable information to evaluate.
Gijs explained that the new platform combines an updated hardware image processor with IQSIGHT’s own image-processing software. In the examples presented, that combination preserved finer detail and more natural-looking images without excessive sharpening, even when the sensor and lens were unchanged.
This helps explain why two cameras with similar resolution specifications can produce different results. Image processing helps determine how much of the captured detail remains useful to an operator or an analytics engine.
Glass entrances, loading docks, and parking garages often contain bright highlights and deep shadows at the same time. High dynamic range processing helps keep both areas visible.
IQSIGHT’s HDR X technology is designed to capture that range while limiting the visual artifacts that can appear around moving objects. Gijs said CPP16 brings the full HDR X capability previously available at 2- and 4-megapixel resolutions to 4K. That gives users seeking higher-resolution coverage another way to retain useful detail in difficult lighting.
Noise reduction can clean up a grainy image, but overly aggressive filtering can also remove details from a moving person or leave a visible trail behind them.
Intelligent Dynamic Noise Reduction, or iDNR 2.5, uses AI-assisted motion detection to distinguish moving subjects from stationary surroundings. It can apply stronger noise filtering to the background while preserving more detail in movement.
Their examples included fine detail disappearing from a moving Ferris wheel and trails appearing behind a walking person with the earlier processing. The newer approach retained more of that detail and reduced the trailing. Removing unnecessary image noise can also reduce the data that must travel across the network and be stored.
The AISP Low Light Enhancer discussed in the session works earlier in the image-processing sequence than iDNR, close to the original sensor data. It uses a larger AI model to reduce noise before later processing amplifies the image.
The intended benefit is a cleaner starting image that can retain useful color at lower light levels. In an early engineering demonstration, a person became lost in the older processing while the enhanced version kept the person visible.
Gijs explained that the additional AI capacity makes it possible to support this demanding processing alongside video analytics. The older hardware would not have enough capacity to deliver the same combination. For an end user, this is a reason to evaluate moving subjects in actual nighttime conditions when comparing cameras.
Infrared illumination helps cameras see at night, but a person close to the lens can receive too much light and appear washed out. Improvements to Intelligent IR are designed to preserve detail in those nearby subjects while maintaining a cleaner image.
Image quality has sometimes differed between fixed cameras and pan-tilt-zoom cameras, commonly called PTZs. Some earlier moving cameras used third-party zoom modules with their own image processing.
IQSIGHT’s in-house zoom modules allow newer PTZ cameras to use its full image-processing approach. Gijs described improvements in color, shadow detail, and handling bright and dark areas, bringing the moving-camera view closer to the quality users expect from the fixed-camera portfolio.
A higher-resolution recording does not automatically mean better AI detection. Matt explained that analytics commonly evaluate a separate image feed at a lower resolution than the main recording. A 4K camera and a 1080p camera can therefore have similar analytics performance if their detection engines receive similar inputs.
The newer platform can analyze higher-resolution inputs, moving from roughly VGA-level images toward 720p or 1080p in the examples discussed. More detail helps the software recognize smaller or more distant objects. The presenters cited up to a 50% increase in detection distance for applicable analytics, depending on the application and setup.
This is edge AI, meaning the analysis runs inside the camera. IQSIGHT also analyzes the image before video compression, giving the analytics access to detail before compression removes information to reduce file size. The practical evaluation should include detection performance at your required distances, with your lens and lighting, alongside recorded image quality.
The IQSIGHT IVA Pro video analytics family supports different operational needs. Matt estimated that people and vehicles account for roughly 90% of the use cases he was describing, making those a useful starting point. The session also covered more specialized capabilities.
| Analytics capability | Practical purpose for the end user |
|---|---|
| IVA Pro Buildings | Detect people and vehicles for common building and property applications. |
| IVA Pro Traffic | Distinguish people, cars, trucks, buses, bicycles, and motorcycles for more specific traffic awareness. |
| IVA Pro Appearance | Describe visible attributes such as clothing colors, sleeve or trouser length, backpacks, and handbags to help narrow a search. |
| IVA Pro Personal Protective Equipment | Recognize supported protective equipment to help monitor defined workplace safety requirements. |
| Visual gun detection | Detect a visible firearm in supported conditions to help operators assess the event and follow their security response procedure. |
Matt also mentioned gender-related appearance classifications. Those are software estimates of visible appearance, not verified personal information. Perimeter detection and license plate recognition address additional needs that deserve their own design considerations.
Too many nuisance alerts can make a perimeter detection system difficult to use. Operators spend time checking harmless activity, and confidence in the alarms declines.
Matt described IVA Pro Perimeter as combining two detection methods. Motion-based analysis can detect distant movement and objects that do not fit a familiar category. AI object recognition can identify trained categories, such as people and vehicles, while helping filter out irrelevant activity.
Each method addresses a weakness of the other. Motion detection can react to things that do not matter. Object recognition can miss an intruder whose clothing, covering, or posture makes them look unlike the people the model learned to recognize. Layering the methods is intended to preserve long-range detection and awareness of unclassified movement while suppressing nuisance alarms, including those associated with animals.
For the end user, the distinctive benefit is the combination of broad detection and selective filtering. It gives the system more than one way to recognize that something deserves attention.
An audience question focused on fence-line detection distance. Matt explained that the lens and the width of the view are central to the answer. As a view becomes wider with distance, a person occupies a smaller portion of the image.
He used a view approximately 300 feet wide as an illustrative point where detection could run out of useful detail in the configurations he was discussing. That was a design example, rather than a universal limit for every camera.
He also described some visible-light lens combinations approaching 1,000 feet and thermal combinations approaching 2,000 feet, while noting that he typically would not design beyond approximately 1,500 feet. In practice, he often divides a perimeter into 500- or 700-foot sections, with overlapping or opposing views to cover the area beneath each camera.
These examples show why an advertised maximum range should be checked against the actual fence line, terrain, lens, and required detection performance. A shorter, well-planned section may provide more dependable coverage than a design stretched to the maximum distance.
Security teams may need to determine which vehicle entered a property, find a vehicle involved in an incident, or manage access to a parking area. IVA Pro License Plate Recognition, paired with vehicle make and model recognition, adds searchable vehicle details to the video.
Capturing attributes such as plate number, vehicle make, model, and color gives an investigation more to work with. Those details become metadata, meaning structured information associated with the video that a compatible system can use for searching or matching.
Gijs explained that the newer platform supports recognition across two adjacent lanes while maintaining full camera functionality, compared with the single-lane configuration discussed for the earlier platform. In the Q&A, he recommended staying within two lanes for reliable results.
For a straightforward parking application, approved and blocked plate lists can be uploaded to the camera using a CSV file. Matt compared the operation to an access control door controller: the camera can continue applying its last saved list if its connection to the management server is interrupted.
Changes to the list still need to reach the camera before they can take effect. Matt also explained that a local match can produce an event without sending out the full stored plate list. The plate decision can remain local, although the operation of connected gates, recording, and other equipment depends on the rest of the system.
For broader investigations, the camera supplies plate and vehicle information to systems such as BVMS video management software. Searches or comparisons involving external databases take place through the appropriate server-side application and integration.
A plate read by itself does not establish who is driving. Linking it with other records is a separate step, and plate histories can still be sensitive information. That distinction helps an organization define the purpose and access rules for its vehicle data.
The presenters confirmed that vehicle make and model recognition information is updated through camera firmware.
Knowing that a person is beside a vehicle does not explain whether anything needs attention. Someone may be tying a shoe, waiting for a ride, or doing something that warrants a closer look.
Matt and Gijs described CPP16 as IQSIGHT’s first platform capable of running generative AI inside the camera without requiring a cloud connection for that processing. Their discussion centered on a visual language model, or VLM, that evaluates imagery and can respond to plain-language prompts.
The model is focused on visual tasks and sized to operate within the camera’s available computing resources. That creates possibilities for custom alerts and interpretation of more complex scenes. It does not establish a person’s intent with certainty, but it can provide another assessment of the visible activity.
One example was debris falling onto a roadway. An operator may need an alert even when nobody can predict whether the object will be a tire or something else. Contextual analysis is intended to help recognize the event without relying entirely on a predefined list of objects.
A second use is alarm verification. An initial detector identifies activity near a fence, then the VLM examines the event to help determine whether it appears relevant. A deer that briefly resembles a person illustrates the kind of event this additional check could help filter.
Matt explained that a second verification stage could allow the first stage to remain more sensitive, reducing the need to tighten its settings so much that relevant activity is missed. This is an additional layer beyond the motion and object-recognition combination discussed for perimeter detection. Its effectiveness should be evaluated against the events and conditions at the site.
Processing within the camera is particularly relevant when an organization cannot send imagery to a cloud service. Some larger AI models still require external computing resources. The on-camera capability discussed in this presentation should also be checked against the exact product, firmware, and release status, especially where a feature was demonstrated as emerging technology.
A security camera is also a networked computer. Matt encouraged end users to think about protecting access to it in much the same way they protect access to a building.
The Secure Element 3.0 discussed in the session provides protected hardware storage for cryptographic keys. Those keys help the device establish trust and protect access to information. The intent is to make sensitive keys difficult to extract, rather than leave them exposed in ordinary device storage.
The presentation also addressed FIPS 140-3 and 4096-bit RSA key support. IQSIGHT identifies the Secure Element in supported products as FIPS 140-3 Level 3 validated. This validation concerns a defined cryptographic module and configuration, rather than an entire installed surveillance system. These details help IT and cybersecurity teams evaluate how the camera protects credentials and participates in a secure deployment. IQSIGHT
For a project with specific security requirements, confirm the applicable certification scope and supported functions for the exact camera and firmware. The benefit of the hardware foundation is a more consistent approach to protecting devices across the deployment.
A security system may need several versions of the same scene: detailed recording for investigations, smooth live video for an operator, a smaller image in a multi-camera display, and a stream suitable for a mobile device.
The presenters described a practical example using a high-resolution recording at 15 frames per second, a full-resolution live view at 30 frames per second, and lower-resolution streams for other uses. CPP16’s increased encoding capacity supports combinations that were constrained on earlier 4K cameras, including two full-frame-rate 4K streams alongside 1080p and 720p streams in the configuration discussed. Recording, live monitoring, and smaller display views can each receive the format they need.
H.265 compression, Intelligent Streaming, and cleaner images from noise reduction help control the amount of video data. Appropriate recording profiles in the video management system can further match storage use to the organization’s actual needs.
Matt urged users to review these settings as storage costs become a greater planning concern. Gijs also discussed further streaming enhancements planned for 2027, with some improvements dependent on processing power that older cameras do not have.
The practical opportunity is to select the image quality and frame rate each task requires. A mobile view and an evidentiary recording can have different needs, and the camera should support both.
Expanding coverage often adds cable runs, mounts, conduit, switch ports, and software licenses. Multi-imager cameras can reduce that infrastructure by placing several independently aimed views in one unit.
The FLEXIDOME dual 7100i IR discussed in the presentation uses two adjustable zoom lenses. It can support 180-degree coverage or watch different directions, such as both sides of a building corner, a staircase, or intersecting hallways. The speakers identified it as a CPP16 camera. IQSIGHT
Four-sensor cameras provide additional options. They can create a 360-degree overview or cover approximately 270 degrees around a corner while another view looks down. Gijs described the CPP15 refresh as adding processing capacity and creating a path to 4K options beyond the four-by-5-megapixel configuration discussed. The exact resolution configuration should be confirmed for the selected product.
The presenters illustrated the infrastructure savings with four fixed cameras and a PTZ consolidated into one modular assembly. In that example, the number of separately installed camera units falls from five to one, an 80% reduction. That figure describes installation count, not an 80% reduction in the total project cost.
IQSIGHT describes the four-imager portion of FLEXIDOME multi+ as using one Ethernet cable, one IP address, and one VMS license. A project should confirm the licensing and integration requirements for its chosen management system and any added camera module. The savings come from the actual configuration, including fewer installation points and supporting accessories. iqsight.com
The FLEXIDOME multi+ 7100i approach adds flexibility to the multi-imager concept. The four-sensor unit can operate on its own, with the option to add a compatible camera module later.
That additional camera can serve several different purposes. A location needing a close view of activity beneath the camera has a different requirement from a vehicle entrance that needs plate detail or a large area that needs automatic tracking.
| Added camera option | Problem it helps address |
|---|---|
| Panoramic fisheye camera | Covers the area beneath the assembly while the four other views focus on more distant areas. |
| High-performance fixed dome | Adds a dedicated view for a specific task, such as license plate recognition. |
| Compatible 5000-series moving camera | Adds PTZ functionality where a more economical moving-camera option meets the requirement. |
| Compatible 7000-series PTZ such as AUTODOME 7100s | Adds closer views and supported tracking capabilities for events identified in the wider overview. |
A panoramic camera can provide broad coverage close to its mounting point, but detail becomes more limited farther away. Pairing it with adjustable multi-imager views lets each part of the assembly serve the distance where it is most useful.
Matt described a traffic example in which the panoramic camera observes the intersection beneath the pole while the four other imagers look farther down the approaching roads. A different configuration could retain an overview while dedicating the added fixed camera to license plates.
This modularity helps an organization match each location to its needs and expand later using supported combinations. It also allows the fifth camera to provide the capability that matters most at that position.
A PTZ can provide valuable detail when it is aimed at the right subject. The difficulty is that activity may begin elsewhere while it is looking in another direction.
MultiGuard PTZ connects the wider multi-imager view with the moving camera. IQSIGHT identifies FLEXIDOME multi+ 7100i paired with AUTODOME 7100s as a supported combination. Any of the four imagers can direct the PTZ toward detected activity. The PTZ can then assess the scene and use its supported tracking functions to follow the subject. IQSIGHT
In attended operation, an operator clicks on a subject in one of the fixed views, directing the PTZ to that location and initiating tracking when configured. In unattended operation, relevant activity can trigger the movement automatically.
By optically zooming in, the PTZ can put more pixels on the subject than the overview provides, even when the overview is recorded in 4K. That produces a closer, more detailed view for assessing or investigating the event while the wider view remains available. The team may have both the surrounding context and a detailed recording of the subject.
The audience asked what happens when another imager sees new activity while the PTZ is already tracking something. Matt described two configurable approaches.
One lets the PTZ complete a tracking period, such as 20 or 30 seconds, before returning to its home position and waiting for another instruction. The other lets a new event interrupt the current tracking task.
The choice depends on the scene. Frequent interruptions in a busy location can prevent the camera from following any event long enough to produce a useful record. A defined tracking period can preserve continuity, while interruption may suit other operational priorities. The configuration should reflect what your operators need to see.
More intelligent cameras can reduce the need to send every analytics task to a central server. When processing occurs at the edge, meaning inside each camera, adding a camera also adds the computing capacity for its supported functions.
Matt contrasted this with a centralized arrangement in which another group of cameras may push an analytics server beyond its available capacity. Distributing suitable work to cameras can make expansion more manageable, particularly as memory and storage costs affect planning. The presenters noted that cameras themselves also contain memory, so those component pressures matter to the devices as well.
They also emphasized that edge and server processing can work together. Some applications still belong on servers, including more extensive database comparisons and other system-wide functions. Recording, retention, and management requirements still need to be sized for the complete deployment.
The rollout described in the October 2026 presentation included the DINION 7100s box, AUTODOME 7100s, and MIC 7100s families. The speakers identified those camera types as available at the time and placed additional FLEXIDOME 8100 and DINION 7100 bullet models on the roadmap for the following January, in 2027.
When reviewing the IQSIGHT camera and video security portfolio, verify the exact model, regional availability, firmware, analytics licenses, and supported combinations. A shared platform creates opportunities across the portfolio, but it does not mean every camera includes every capability covered in the presentation.
The most useful demonstration begins with a problem your team already encounters. That could be a person disappearing into a dark part of a parking lot, repeated perimeter nuisance alarms, an entrance that needs reliable plate recognition, or a PTZ that misses the beginning of an incident.
Ask to see that situation reproduced with the proposed camera, lens, analytics, and management software through a proof-of-concept evaluation. Include the lighting and distances that challenge your current system. Review the resulting alerts, recorded video, and operator workflow together.
Matt and Gijs encouraged attendees to test the technology firsthand. Their session pointed to demonstrations of analytics at the edge, the emerging VLM capabilities, modular FLEXIDOME multi+ configurations, and BVMS features such as real-time person tracking. That same practical approach belongs in a project evaluation.
Contact the MidChes team to discuss a demonstration around your site’s requirements. Seeing how these capabilities handle your actual security challenges will make it easier to decide which improvements belong in your next system upgrade.
Traditional cameras primarily record video for live monitoring and investigations. AI-enabled security cameras can also recognize people, vehicles, and specific activities, helping security teams identify potential threats and respond more efficiently. IQSIGHT combines these capabilities with advanced image processing and analytics that operate directly within supported cameras.
IQSIGHT IVA Pro Perimeter combines motion-based detection with AI object recognition. This helps distinguish potentially important activity from nuisance events caused by animals or other movement. IQSIGHT also demonstrated emerging visual AI technology that can provide an additional layer of alarm verification before an alert reaches an operator.
Yes, although performance depends on available lighting, camera placement, and the selected model. Technologies such as HDR X, Intelligent Dynamic Noise Reduction, AISP Low Light Enhancer, and Intelligent IR help preserve image detail, reduce noise, and improve visibility in challenging lighting conditions.
Not necessarily. IQSIGHT performs many supported video analytics functions directly inside its cameras, reducing dependence on cloud processing. Its newer CPP16 platform is also capable of supporting on-camera generative AI. Specific AI capabilities and availability depend on the camera model, firmware, and application.
No. Recording resolution and AI detection performance are not necessarily the same. IQSIGHT's newer processing platform can analyze higher-resolution images, improving the ability of supported analytics to recognize smaller or more distant objects. The presentation cited detection distance improvements of up to 50% in applicable configurations.
In some applications, yes. IQSIGHT multi-imager cameras combine multiple adjustable views into one device, reducing the need for separate camera installations, cabling, and mounting hardware. The modular FLEXIDOME multi+ can also incorporate additional camera capabilities, including panoramic viewing, license plate recognition, or PTZ tracking. Actual coverage and savings depend on the site.
IQSIGHT MultiGuard PTZ allows fixed camera views to direct a compatible PTZ camera toward detected activity. The PTZ can zoom in and automatically follow a subject while the fixed cameras continue providing broader coverage. This helps capture important details without requiring an operator to manually locate every event.
Supported IQSIGHT cameras can recognize license plates and vehicle attributes, including make and model. On-camera approved and blocked plate lists can support basic vehicle access decisions. A camera can continue using its last saved list if its connection to the video management server is interrupted, although updates and connected equipment may depend on the wider system.
They can help. IQSIGHT combines H.265 compression, Intelligent Streaming, and image noise reduction to manage the amount of video data generated. Organizations can also configure recording quality and frame rates for different needs, potentially reducing storage and network demands without unnecessarily sacrificing important video detail.
IQSIGHT's newer camera platform incorporates Secure Element 3.0 technology to protect sensitive cryptographic keys within the device. Supported products also incorporate cryptographic capabilities designed to meet demanding security requirements. These protections complement proper network configuration, access management, and firmware maintenance.
Start with the security problems you need to solve, whether they involve poor nighttime visibility, excessive false alarms, perimeter coverage, vehicle monitoring, or escalating storage costs. A site-specific demonstration or proof of concept can help determine which camera designs, analytics capabilities, and configurations will deliver measurable operational benefits.