Axxonsoft has introduced a neural network filter for Object Tracker that analyses moving or abandoned objects to filter out false alarms from situation analysis detection tools. The filter precisely recognises specified types of objects, e.g., humans or vehicles. To guarantee high-quality filtering, the neural network learns to perform customer-specific tasks from on-site recorded videos.

Users can now simply replace the neural network file for the fire / smoke detection tools with the updated version. The update enables you to request additional network deep learning from Axxonsoft and receive a ready-to-use neural filter file.

A false alarm filter for Facial Recognition has also been added that filters out objects that are most likely incorrectly identified as faces. False recognition results will appear on the Events Board but will be ignored during searches in the video footage archive. Facial Recognition now recognises age and gender. Recognition data can be displayed on the Events Board and saved in the database for further statistical analysis.

The new version also incorporates a Large Number of Objects detection tool which triggers an alarm upon reaching a specified number of objects within a specified area. This latest update also introduced tracking for the frame around a vehicle license plate. In addition, the company has reduced the CPU load needed to recognise excessively large license plates.

In addition to Object Tracker, the basic Video Motion Detection tool now generates metadata that can be used for Momentquest forensic search. Since the VMD requires much less computing power, the cost of the system can be lowered by using more cameras per server.

Users can now perform facial and vehicle number searches across all cameras within an Axxon domain. Amongst many other new features, archives with video footage can now also be automatically combined for playback.