Situational Analytics and Event Search

They register object movement in the frame: motion and stopping, prolonged presence, entering and leaving a defined area, moving from one zone to another, line crossing, presence of one or multiple objects, abandoned items. For more precise detection, you can configure object color, speed, type and size accounting for perspective.

The analytics use a scene description generated by a motion detector, a classic object tracker, or a neurotracker. They differ in accuracy, resource intensity and configuration options.

Neurotracker is used together with situation analysis detectors to detect and track objects of a certain type. It significantly reduces the number of false alarms caused by visual noise: vegetation in the wind, ripples on water, fog and precipitation, shadows and glare, movement of objects that are not of interest.

ITV software has built-in neural networks for detecting people and vehicles. We also offer neural network training for recognizing other object types and operating in difficult conditions.

Counts objects in a selected area of the frame. Alerts when the number of objects becomes greater than or equal to (or less than or equal to) a specified value, or when the number of objects changes. Like the neurotracker, the neurocounter can use the built-in neural networks for counting people and vehicles, or specially trained neural networks for counting other types of objects.

Detects the presence of specific objects using a neural network. Unlike the neurotracker, this detector does not track object movement and therefore consumes fewer computing resources. A neural network for recognizing a given object type can be trained independently or with the involvement of ITV specialists.

To quickly find recordings, simply select an area of the frame, and the system will find all events related to movement in that area. You can also set more precise criteria, configured similarly to the situation analysis detectors — for example, line crossing and object size. The search uses recorded metadata generated by a motion detector, a classic tracker, or a neurotracker.