Similar to human brains, the more AI-powered devices "learn" a subject, the better they analyze and respond to it. Hikvision provides methods for its deep-learning algorithms to continually improve their accuracy in any environment where deployed.


As the algorithms evolve, they gain a better understanding of objects in a scene and higher efficiency in problem-solving there.

Learning from users' feedback


Hikvision’s AcuSense technology, for example, is enhanced with the Learn-by-Example mode of Self-Learning Analytics. In this mode, the algorithms learn with feedback from users on alarm triggers. Users mark the false alarms generated within a particular scene and from those examples the algorithms will learn to minimize similar alarm errors in follow-up trials.

Fully automated learning


The Learn-by-Self mode enables fully automated learning by Hikvision algorithms, embedding an inference engine in the back-end device to automatically optimize algorithm accuracy in fixed cycles. This mode requires virtually no effort from users and will be available on Hikvision's video recorders in the near future.

Learn more about Hikvision AcuSense Technology