The new version of intuVision VA introduces a comprehensive library of pre-trained models for object detection in video, complementing our proven pixel-based detection and user-trained object classification. Users now have the flexibility to select the optimal detection method for each camera view and application, ensuring accurate results across a wide range of requirements.

All intuVision VA modules: Security, Traffic, Parking, Retail, Manufacturing, Face and Text Detection now offer two complementary alterantives - providing the most reliable detection and tracking, no matter the challenges present in camera views and project requirements.

Our patented Pixel-Based Object Detection is a flexible method of tracking active objects in a scene, which allows for detection of any moving object, regardless of its type/class. While using this method, intuVision VA provides in-situ classification model training, requiring just few samples, in limitless user selected categories, from people to fish, across many application areas. intuVision VA v.18, introduces the alternate method of detection using multitude of Pretrained Object Models, expanding the reach of intuVision VA to more challenging scenes and environments.

The model-based detection method exhibits particularly superior performance for person and vehicle counting in low camera views and in scenes where cast shadows interfere with the detection. Pixel-based detection method shines in environments where the object type is not known in advanced, in custom applications, and in typical top-down camera views.

The video below highlights the differences between the two detection methods and showcases scenes and application areas most suitable for each.