The Avision™ core scalable video analytics engine
Artivision's core proprietary video content analysis (VCA) engine, Avision™, detects predefined scenarios and events over a stream of video images and generates transactional data analysis of frequency and behavior. Avision™ is based on an object-based algorithm platform utilizing advanced Temporal Analysis concepts, and uses artificial intelligence to enable self-learning of real-world environments and scenes.
Based on this core video analytics technology, Artivision has developed a comprehensive range of analytics applications for detecting and tracking various incidents, behaviors and individuals. Furthermore, Avision™ serves as the foundation framework for flexible applications in diverse industries, consisting of a hierarchy of algorithms and tools, including object detection and tracking, change detection between images, still image segmentation, moving object segmentation, MPEG block segmentation, and many others. These algorithms form the core technology for many advanced artificial-intelligence applications for understanding and recognizing contents of images, video and media.
Analyze objects, not pixels
The Avision™ video content analysis engine learns a scene’s background and segments each object in the frame. This enables to analyze and track objects instead of relying on mere pixel changes between frames, as commonly used in simplistic VMD (Video Motion Detection). The ability to understand object behavior leads to less false detections, especially in outdoor environments where ambient image changes arising from lighting, shadows, vegetation and weather are common.
Self-learning with minimal settings and maintenance
After a short and straightforward initial setting, the Avision™ video content analysis engine autonomously optimizes its sensitivity and configuration to each scene, requiring minimal fine-tuning and ongoing field maintenance.
Robust performance with advanced false alarm filters
The Avision™ video content analysis engine enables robust performance even under harsh weather and environmental conditions, using advanced object tracking, ambient motion cancellation and other false alarm filters. The analytics engine automatically adjusts to changes in the scene, effectively differentiating between environmental “noise” and true events.
Apply depth of perspective and aspect ratio – effective detection in 3 dimensions.
Ignore blinking lights and insects – automatically adjust for broad range of illumination (sun rise and set, switching lights on/off, shadows, reflections), while cancelling nuisances such as insects or birds.
Identify even slow motion – identify extremely slow moving objects (such as crawling people).
Cancel vibrations – compensate for vibration movement in fixed video cameras, usually when positioned on poles, or deployed near railway lines, highways and waterlines.
Filter directional movement – focus detection on one or more specific directions.
Anti-tampering protection – verify that the camera has not shifted its position, and alert for any tampering attempts.