Combine high-quality video monitoring, AI-powered detection, and innovative surveillance technology to deliver the ultimate protection for your customers.
Fluendo’s expertise spans video capture, analysis, transmission, and secure storage for later review, making us a trusted technological partner for improving and developing advanced video surveillance solutions. With deep knowledge of major multimedia frameworks, we will help you boost your video security software with the latest state-of-the-art technologies.
Fluendo’s expertise spans video capture, analysis, transmission, and secure storage for later review, making us a trusted technological partner for improving and developing advanced video surveillance solutions. With deep knowledge of major multimedia frameworks, we will help you boost your video security software with the latest state-of-the-art technologies.
No matter which operating system or multimedia framework you use, Fluendo’s expertise can help you reduce time-to-market for your CCTV monitoring software.
Rely on production-ready components that provide stability to your surveillance software, improving your capturing and streaming processes. No more hard resets to your service.
Improve your reliability by having proper validation systems that can emulate your video surveillance AI as a whole with long-running and regression tests.
Bring the latest codecs into your newly created media files to reduce your storage size.
No matter which operating system or multimedia framework you use, Fluendo’s expertise can help you reduce time-to-market for your CCTV monitoring software.
Rely on production-ready components that provide stability to your surveillance software, improving your capturing and streaming processes. No more hard resets to your service.
Improve your reliability by having proper validation systems that can emulate your video surveillance AI as a whole with long-running and regression tests.
Bring the latest codecs into your newly created media files to reduce your storage size.
our case studies
Developments that bring real-world results, these case studies show how our solutions help your business achieve goals and enhance user experiences.
The client was transitioning from a legacy Windows-only application to a modern, cross-platform solution using GStreamer, targeting Windows, macOS, and Linux. The application, built with a C# front end and a C/C++ backend layer, suffered from instability, including memory leaks and frequent crashes.
We were brought in mid-development to help bring the project back on track. Our involvement focused initially on stabilizing the system, followed by the implementation of new features such as masking and video wrapping. Leveraging our GStreamer expertise, we explored and tested multiple architectural approaches, ultimately delivering a stable and functional application in time for the product’s launch.
We provided GStreamer expertise to support the development of a native Windows application built on the GStreamer framework. This included a major codebase refactor to resolve critical bugs, enhance performance, and improve overall system stability. To ensure long-term success, we also delivered tailored GStreamer training to the client’s team, empowering them to maintain and evolve the application independently.
For a deeper look at how we transformed a Windows-only legacy app into a modern, cross-platform GStreamer solution, check out our article Revolutionizing video management: GStreamer project on Windows, where we share insights that echo the work delivered in this project.
The client, a global provider of hardware and software for video surveillance systems, leverages AI to analyze video streams from multiple IP cameras. These streams are then aggregated, encoded, and delivered in real time through their client-facing web application.
As they transitioned from an FFmpeg-based solution to GStreamer, the client engaged us to perform an in-depth analysis of their new streaming pipelines. Leveraging our GStreamer expertise, we identified key areas for optimization and enhanced their pipelines to take full advantage of NVIDIA® hardware acceleration for improved performance and efficiency.
We conducted a thorough analysis of the client’s GStreamer pipelines and underlying source code. Based on our findings, we proposed a revised architecture designed to enable zero-copy processing, leverage NVIDIA® hardware acceleration, and ensure seamless integration with the WebRTC streaming protocol.
The client needed a robust video management system (VMS) application to support real-time security monitoring and playback of surveillance footage from various locations—on-site, centralized, and mobile. The system architecture consisted of a C#-based graphical user interface (GUI) and a C/C++ multimedia backend, with a clear separation between the two components.
To support maintainability and catch issues early in the development cycle, the client required a reliable testing strategy for the backend. Rather than traditional unit tests—which were not suitable for this phase—we proposed and implemented a suite of functional automated tests using the backend’s JSON API. These tests were designed to integrate with the client’s CI/CD pipeline, enabling early detection of bugs and improving development efficiency.
We established a streamlined development workflow focused on quality and reliability. As part of the solution, we integrated automated functional testing into the continuous integration process. This ensured consistent validation of the multimedia backend and enabled the generation of clear, comprehensive reports. These reports made it easy for the client to track progress, identify issues early, and maintain confidence in the stability of the system throughout development.
If you want to learn more about how automated testing strengthens multimedia applications, take a look at our article Integrating automated testing in multimedia, where we share best practices and insights that complement the work done in this project.
Fraunhofer HHI required the integration of Digitally Signed Content (DSC) support into GStreamer to ensure content integrity and traceability across multimedia workflows. At the time, GStreamer did not provide standardized mechanisms for embedding and validating digitally signed metadata within encoded video streams, which created a gap for organizations seeking trusted and verifiable media pipelines.
Fluendo’s expertise was essential because of its 20-year history as a leader in the GStreamer community and its previous success with the SPIRIT OC1 project, which served as the baseline for this engagement. As multimedia experts with deep insight into the framework, the team was uniquely positioned to implement DSC SEI messages on top of H.266/VVC bitstreams. This project aimed to deliver a modular, maintainable solution that could be contributed upstream to the open-source community.
The diagnostic process identified that traditional authentication methods relied on container-level signatures, which were easily stripped or manipulated. To address this, the solution focused on embedding cryptographic signatures directly into the video bitstream at the Network Abstraction Layer (NAL) unit level. This ensured that authentication information traveled with the video data itself, making it significantly harder to tamper with.
The implemented solution involved extending the H.266 multimedia parsers to extract SEI messages carrying DSC metadata and expose them as GstMeta. New GStreamer elements were developed to handle DSC signing and verification, allowing for the generation of signing metadata in compliance with JVET standards and the validation of content using public keys. These elements emitted signals to notify the system of validation success or failure, providing a robust framework for real-time media authentication.
The engagement resulted in a modular implementation that was contributed upstream to the GStreamer community, ensuring long-term maintainability and industry adoption. Further technical details were shared in a related blog post.
Ensured full alignment with JVET-AK0194 and ITU H.274 (VSEI) standards for video authentication.
Enabled trustworthy signing and authentication of coded video data at the elementary stream level.
Developed encapsulated GStreamer elements for signing, inserting, parsing, and verifying metadata.
Ensured full alignment with JVET-AK0194 and ITU H.274 (VSEI) standards for video authentication.
Enabled trustworthy signing and authentication of coded video data at the elementary stream level.
Developed encapsulated GStreamer elements for signing, inserting, parsing, and verifying metadata.
our use cases
These use cases present conceptual examples of how our ideas and technologies could address real-world industry challenges.
In high-risk environments such as elderly care facilities, hospitals, public transit stations, or workplace safety zones, detecting human falls instantly is essential for preventing serious injuries and enabling rapid emergency response. However, continuous manual video surveillance monitoring is resource-intensive and prone to oversight.
A growing need exists for automated fall detection systems that operate in real-time, are accurate, and respect user privacy. Such a system would assist caregivers, security teams, or emergency responders by automatically identifying falls and triggering alerts, without relying on wearable sensors.
With advancements in AI pose estimation systems, a computer vision–driven solution could be deployed directly on-site to provide privacy-compliant, accurate, and low-latency fall detection.
Our fall detection application uses a deep learning model to estimate body key points and track motion trajectories frame-by-frame. By analizing changes in body angles, joint angles and movement velocity, the system can accurately detect genuine falls while filtering out benign movements (e.g., sitting or bending). This fall detection solution is designed for seamless integration and works with standard CCTV and IP camera streams using a GStreamer pipeline.
Enables immediate response to incidents, reducing injury severity and improving care quality.
Easy plug-and-play integration with existing systems, combined with configurability, enabling quick deployment and customization to meet diverse environment requirements.
Processes data locally, without needing personal identity recognition—ideal for GDPR-sensitive environments.
Enables immediate response to incidents, reducing injury severity and improving care quality.
Easy plug-and-play integration with existing systems, combined with configurability, enabling quick deployment and customization to meet diverse environment requirements.
Processes data locally, without needing personal identity recognition—ideal for GDPR-sensitive environments.
Public spaces such as transportation hubs, university campuses, shopping centres, and event venues face growing security risks from both concealed and openly carried knives. These threats put heavy pressure on security teams to detect weapons quickly and respond before incidents escalate. Traditional manual monitoring of multiple CCTV or IP camera feeds is labour-intensive and prone to human error and delayed reactions, particularly in crowded or fast-moving environments.
An AI-powered, real-time knife detection system can significantly improve situational awareness by automatically scanning live video streams for potential weapons and sending instant alerts to security personnel. Thanks to recent advances in computer vision and low-latency edge computing, these intelligent detection models can now be seamlessly integrated into existing surveillance networks without costly infrastructure changes.
Our advanced deep learning object detection models are trained on diverse datasets that cover varying lighting conditions, camera angles, and crowd densities. This enables the system to recognize knives accurately in live and recorded video streams, even in challenging environments.
Integrated into the GStreamer multimedia framework, the AI pipeline continuously processes incoming CCTV or IP camera feeds in real time, delivering high detection accuracy with low false-positive rates. The result is a reliable, scalable, and proactive security tool that enhances public safety by identifying potential threats as they happen.
Identifies potential threats early, allowing security teams to intervene before incidents escalate. This proactive approach reduces risks by enabling swift response in high-traffic, dynamic environments, enhancing overall safety.
Trained on diverse scenarios, our AI-driven system minimizes false positives while maintaining high sensitivity. It ensures accurate, reliable threat detection across varying conditions, providing security teams with precise alerts in dynamic environments.
Processes data locally for faster performance and enhanced data security. This approach ensures quick response times while maintaining privacy and compliance, making it ideal for real-time threat detection in dynamic and sensitive environments.
Identifies potential threats early, allowing security teams to intervene before incidents escalate. This proactive approach reduces risks by enabling swift response in high-traffic, dynamic environments, enhancing overall safety.
Trained on diverse scenarios, our AI-driven system minimizes false positives while maintaining high sensitivity. It ensures accurate, reliable threat detection across varying conditions, providing security teams with precise alerts in dynamic environments.
Processes data locally for faster performance and enhanced data security. This approach ensures quick response times while maintaining privacy and compliance, making it ideal for real-time threat detection in dynamic and sensitive environments.
In modern live broadcasting and video surveillance, operators struggle to keep moving subjects framed when using traditional PTZ (Pan-Tilt-Zoom) cameras in dynamic environments such as lecture halls, theaters, or public spaces. An AI-driven subject tracking solution can autonomously detect and follow speakers, performers, or persons of interest, reducing reliance on manual control and ensuring consistent coverage.
By integrating real-time subject tracking with PTZ hardware, this technology maintains consistent coverage even when subjects move unpredictably. Recent breakthroughs in low-latency edge computing make it possible to deploy this system on-site for both live event production and security surveillance, delivering precise and reliable tracking without additional operators.
Recent low-latency edge computing advancements make precise subject tracking feasible for both live production and surveillance.
Our smart PTZ camera control solution combines real-time multi-target tracking with precise pan, tilt, and zoom control to keep your subjects in view at all times. Integrated via a GStreamer pipeline and deployed directly on-site, the system processes live video feeds to detect and assign bounding boxes to selected targets.
A bidirectional PTZ interface translates bounding box movement into smooth pan, tilt, and zoom commands, ensuring seamless tracking without abrupt or jerky transitions. Equipped with re-identification capabilities, the tracking module can handle occlusion and rapid movement and re-acquire targets even after they leave and re-enter the frame.
Minimizes manual intervention and streamlines operations, enabling more efficient workflows and reducing the need for dedicated resources in live production or surveillance.
Delivers precise, professional-level shot composition with high-quality tracking of moving subjects, ensuring optimal visual clarity in live events and surveillance.
Enables efficient, on-site tracking with minimal infrastructure, offering customizable solutions for mobile rigs and remote setups.
Minimizes manual intervention and streamlines operations, enabling more efficient workflows and reducing the need for dedicated resources in live production or surveillance.
Delivers precise, professional-level shot composition with high-quality tracking of moving subjects, ensuring optimal visual clarity in live events and surveillance.
Enables efficient, on-site tracking with minimal infrastructure, offering customizable solutions for mobile rigs and remote setups.
In retail, logistics, and critical infrastructure security sectors, detecting suspicious or abnormal behavior quickly is essential for preventing losses and mitigating risks A common risk scenario is when individuals loiter or linger in restricted or high-value areas (e.g., product aisles, exits, stockrooms, or sensitive perimeters) beyond normal dwell times, behavior that may indicate theft, intrusion, or a developing security threat. Traditional methods rely on manual CCTV monitoring or basic motion alerts, which are often labor-intensive, prone to false positives, and slow to respond.
AI-powered computer vision now enables a more robust approach by combining person detection, tracking, and region-of-interest (ROI) analysis to identify individuals who remain in defined zones longer than expected—automatically flagging potential threats or anomalies in real time.
The solution leverages a GStreamer-based video pipeline integrated with AI models for person detection and multi-object tracking (MOT). Security teams can define Regions of Interest (ROIs) within each camera view, such as entrances, exits, restricted zones, or high-value product aisles.
The system continuously calculates dwell time within each ROI for every detected individual. When time exceeds a configurable threshold, the system automatically triggers an alert or logs the event for review, enabling early detection of loitering, intrusion, or potential marauding behaviors.
Designed for real-time edge deployment, the system minimizes latency. It reduces bandwidth requirements while supporting centralized deployments for large-scale environments such as retail chains, logistics hubs, and critical infrastructure facilities.
Proactively detects behavior patterns associated with theft, loitering, or tampering before incidents occur.
By lowering dependence on manual surveillance, organizations can reduce operational expenses.
Enhances the perception of safety and control in stores, warehouses, or public venues—reinforcing customer and employee confidence.
Proactively detects behavior patterns associated with theft, loitering, or tampering before incidents occur.
By lowering dependence on manual surveillance, organizations can reduce operational expenses.
Enhances the perception of safety and control in stores, warehouses, or public venues—reinforcing customer and employee confidence.
In high-traffic environments such as retail stores, transportation hubs, entertainment venues, and smart buildings, understanding real-time occupancy and pedestrian flow is critical for safety compliance, operational efficiency, and business intelligence. However, traditional methods, manual counting, turnstile-based systems, or basic sensor arrays tend to be costly, inaccurate, and unable to provide the spatial flow insights needed for informed decision-making.
A growing need exists for automated, camera-based solutions that deliver accurate occupancy counts, directional flow analysis, and zone-level breakdowns, all in real-time, without requiring personal identification, and in compliance with privacy regulations such as GDPR. With today’s advances in computer vision and AI, it would be possible to deploy intelligent video analytics systems that transform existing surveillance infrastructure into a powerful source of actionable data, processed entirely at the edge.
A people-counting and flow-analysis application would use deep learning–based detection and multi-object tracking models to identify and track individuals across video frames in real-time. With virtual lines and zones defined in the camera’s field of view, the system would count entries, exits, and crossings per area, while tracking movement direction and dwell time.
The solution would be built as a modular GStreamer pipeline using Fluendo AI Plugins and the AI Engine:
People Detection — A real-time object detection model would identify people in each frame, even in crowded or partially occluded scenes, running inference directly on the GPU through Raven’s fully hardware-accelerated pipeline.
Multi-Object Tracking — Detected individuals would be assigned persistent IDs and tracked across consecutive frames using the tracking module, enriched with deep learning–based re-identification capabilities.
Flow & Direction Analysis — The system would calculate the direction of movement for each tracked individual using trailing metadata, enabling directional flow maps, heatmaps, and dwell-time analytics.
This pipeline would integrate seamlessly with standard CCTV and IP camera streams and could be optimized for edge computing deployment through the AI Engine, enabling real-time processing on local hardware without cloud dependency. Full hardware acceleration across Intel, AMD, and NVIDIA GPUs would ensure efficient performance even in resource-constrained environments.
Automated camera-based counting would eliminate the need for manual staff, turnstiles, or dedicated sensor hardware, significantly reducing operational costs.
Real-time occupancy alerts and zone-level monitoring would enable proactive crowd management, reducing the risk of safety incidents and liability exposure.
All AI processing would run locally on edge devices without facial recognition or personal data storage, ensuring GDPR and capacity regulation compliance by design.
AI-powered detection, tracking, and flow analysis on edge devices would deliver capabilities beyond traditional counting methods.
Automated camera-based counting would eliminate the need for manual staff, turnstiles, or dedicated sensor hardware, significantly reducing operational costs.
Real-time occupancy alerts and zone-level monitoring would enable proactive crowd management, reducing the risk of safety incidents and liability exposure.
All AI processing would run locally on edge devices without facial recognition or personal data storage, ensuring GDPR and capacity regulation compliance by design.
AI-powered detection, tracking, and flow analysis on edge devices would deliver capabilities beyond traditional counting methods.
Bits & Bytes
Explore our blog, one byte at a time. Our team unpack our latest news, industry insights and in-depth articles to connect you with the multimedia world.Blog
Read more about our work
Securing Video Pipelines: Digitally Signed Content (DSC) Support in GStreamer
broadcasting, software-security, video-surveillance, codecs, gstreamer, outsourceImplementing cryptographic video authentication in GStreamer to combat deepfakes and ensure content integrity.
Real-time 4K face anonymization benchmark (GStreamer plugin)
broadcasting, video-surveillance, automotive, multimedia-edge-ai, gstreamer, events, fluendo-ai-plugins, anonymizer, ravenPerformance benchmark analysis of real-time 4K AI video anonymization.
Privacy-preserving AI video surveillance with edge intelligence
video-surveillance, multimedia-edge-ai, gstreamer, events, fluendo-ai-plugins, ravenGDPR-compliant AI video surveillance with real-time edge intelligence.
Image superresolution with GStreamer
broadcasting, video-surveillance, sports, multimedia-edge-ai, gstreamer, events, fluendo-ai-plugins, ravenAI-powered 4x video upscaling with Fluendo's GStreamer plugin.