Xeoma for Mac

• New “Timer” module: Added a new filter module “Timer” that allows setting the duration for which subsequent modules in the chain will operate, along with break intervals between these working cycles. Unlike the “Scheduler” module, the “Timer” module works exclusively with time and allows setting cyclic operation intervals with infinite repeats, specifying how long the module should work and how long it should pause. Different cameras can be configured with different working cycles and break intervals. Additionally, to conserve bandwidth, the module can stop receiving video streams from cameras during break periods. The module is available in Xeoma Standard and Xeoma Pro editions. Read more about the module
• New “Duration Detector” module: Added a new filter module “Duration Detector” that triggers when an event from the previous detector lasts longer than the specified time. The module can also work on the opposite principle and react, conversely, to the absence of triggers for the specified duration or longer. With the “Duration Detector”, you can respond to an event that lasts too long or, vice versa, react to the absence of target objects or events in the frame for longer than expected. The module is available in Xeoma Standard and Xeoma Pro editions. Read more about the module
• New “Inverter” module: Added a new filter module “Inverter” which automatically reverses the state of the previous module in the chain (or several modules if they are connected to the “Inverter” in parallel). Thus, if the previous module is triggered and passes the signal, the “Inverter” will block it. And vice versa: if the previous module is not triggered, the “Inverter” will pass the trigger signal to the next module in the chain. The module is available in Xeoma Standard and Xeoma Pro editions. Read more about the module
• Face Recognition: For all recognition methods, we’ve improved the recognition of faces at angles relative to the camera — the so-called background “face rotation adjustment”. It was implemented because creating ideal conditions for face recognition is challenging: usually cameras are mounted high, and people often turn away. Therefore, the new feature will help analytical algorithms automatically perceive faces as if the person were looking directly at the camera.
• Face Recognition: Added automatic retraining when switching recognition methods. Now, if you’ve trained a face database for one recognition method, you can switch to another recognition method, and the program will automatically retrain the new algorithm using using previously added faces. Previously, different recognition methods required their own separate face databases.
• Face Recognition: The ‘Statistical analysis’ recognition algorithm has been removed, the ‘Artificial Intelligence’ recognition algorithm has been renamed to ‘Previous generation 1.0’, and the ‘Artificial Intelligence with CNN’ – to ‘Previous generation with CNN’. The licensing terms for both remaining options remain unchanged: a license for the Additional module based on the number of faces in the database is required.

Full list of changes available here