Visidon Demos at EVS 2026
At this year’s event, we are presenting our portfolio of AI-powered video enhancement solutions, built for real-time deployment on embedded platforms and designed to tackle key challenges in low-light imaging, long-range capture, and perceptual video quality.
Live Demos at Our Booth
AI Low-Light Enhancement (RAW Domain, Real-Time on Qualcomm)
Our core embedded video technology operates directly in the RAW domain, enabling superior restoration before ISP processing. Running in real time on a Qualcomm platform, it delivers clean, sharp video with accurate colors—even in extremely low-light conditions.
To demonstrate its capabilities, we’re using a lightbox setup with minimal lighting, specifically designed to highlight performance in ultra low-light and global motion scenarios.
Real-Time Video Face Super Resolution for conference room cameras
This solution uses a dedicated deep learning model trained for facial reconstruction. It enhances facial regions frame by frame, restoring fine details while maintaining identity consistency and temporal stability.The result is clear, natural-looking faces even at long distances, making it ideal for video conferencing and multi-person scenes.
This particular live demo is running on the Qualcomm QCS8550 Hexagon NPU and performs selective enhancement only on detected face regions of interest, leaving the rest of the frame unchanged for maximum efficiency and natural scene preservation. The model is fully quantized for the QCS8550 NPU and is designed to operate within the camera pipeline’s strict 33 ms frame budget.
The implementation supports up to 4 concurrent face tiles per frame with approximately 14 ms inference latency, while sustaining 30 fps operation alongside the ISP and video encoder workloads. This enables practical deployment in real-time conferencing cameras and edge AI video systems without compromising overall system performance.
Let us know if you are interested in production integration support, reference design guidance, or an Android demo APK for evaluation. You can also read more about the product from our previous blog here.
Video Face Super Resolution live demo 2026
Visidon Traffic Image Booster
Designed for intelligent transportation and surveillance, this lightweight enhancement module operates on processed video streams, making it fundamentally different from our on-device RAW-domain solutions. It improves visibility in degraded conditions such as darkness, weather interference, and long-range capture—without requiring changes to the camera pipeline.
By enhancing image quality at the source, it significantly boosts downstream analytics like object detection and traffic monitoring. It integrates seamlessly into existing pipelines and is optimized for platforms such as NVIDIA Tegra.
Contact our sales team on-site!
Vaida Jasulaityte
Business Development director
Tommi Määttä
Head of Imaging & AI Technologies