text over image889
  • What is Hikvision’s Self-Learning Analytics technology?

  • There is no one-size-fits-all solution for video analytics in perimeter protection. Every deployment environment is unique, and there may be objects that are not covered in the training samples. Hikvision has developed a technology that enables an algorithm to learn and improve in the actual deployed environment.

  • What is Hikvision’s Self-Learning Analytics technology?

     

    There is no one-size-fits-all solution for video analytics in perimeter protection. Every deployment environment is unique, and there may be objects that are not covered in the training samples. Hikvision has developed a technology that enables an algorithm to learn and improve in the actual deployed environment.

text over image168
  • The technology, Self-Learning Analytics, features the self-supervised learning (SSL) paradigm in machine learning. It doesn't rely on manual labeling, enabling video analytics appliances to adapt to the dynamics within the scene autonomously.

    Self-Learning Analytics incorporates an inference engine in an NVR to automatically identify and annotate false alarms generated in the initial round of analysis. It periodically takes that information to train and deploy a new algorithm in the NVR.

    Over time, the algorithm becomes better at distinguishing between people, vehicles, and other objects, especially those distinct objects in the deployment environment.

  • The technology, Self-Learning Analytics, features the self-supervised learning (SSL) paradigm in machine learning. It doesn't rely on manual labeling, enabling video analytics appliances to adapt to the dynamics within the scene autonomously.

    Self-Learning Analytics incorporates an inference engine in an NVR to automatically identify and annotate false alarms generated in the initial round of analysis. It periodically takes that information to train and deploy a new algorithm in the NVR.

    Over time, the algorithm becomes better at distinguishing between people, vehicles, and other objects, especially those distinct objects in the deployment environment.

How does Self-learning Analytics works?

text over image1514
  • The first analysis 

  • (Tagged by persons, vehicles, or others)

  • Sample Collection

  • The second analysis 

  • Self-iteration

  • Self-annotate

  • Self-training

  • Alarm accuracy improved

  • Inference process

    Secondary analysis of all the reported images

  • Training process

    It uses samples from specific scenario conditions to train and deploy new algorithms

  • Inference and training work simultaneously in NVR, iterating algorithms according to scenarios.

text over image1406
  • The power of Self-Learning Analytics 

  • Deploy, learn, and evolve

  • The algorithm iterates automatically after the deployment, with no effort needed from the user.

  • Applicable to all scenarios

  • The algorithm adapts to the particular conditions of a scenario with growing detection accuracy.

  • Usable in all video channels

  • Running in the NVR, the feature supports application across all connected video channels and is compatible with conventional non-AI cameras.

  • Deploy, learn, and evolve

  • The algorithm iterates automatically after the deployment, with no effort needed from the user.

  • Applicable to all scenarios

  • The algorithm adapts to the particular conditions of a scenario with growing detection accuracy.

  • Usable in all video channels

  • Running in the NVR, the feature supports application across all connected video channels and is compatible with conventional non-AI cameras.

Where to use Self-Learning Analytics?
Residences
Warehouses
Parking Lots
Business Parks

Which products come equipped with Self-Learning Analytics? 

The Self-Learning Analytics technology is available on Hikvision’s VPro Series NVRs.

Recommended Product Solutions

Perimeter protection with self-learning

  • Up to 32-ch
  • Up to 64-ch
text over image1517
  • Network Camera with AcuSense

  • K-VPro Series NVRs

  • · Core value

    - Self-iterative algorithms to improve alarm accuracy

    - Reduce manual maintenance costs to save labor and effort

  • · Application scenarios

  • Residences

  • Warehouses

  • Network Camera with AcuSense
  • K-VPro Series NVRs
  • · Core value

     
    - Self-iterative algorithms to improve alarm accuracy
    - Reduce manual maintenance costs to save labor and effort
  • · Application scenarios

  • Residences
  • Warehouses
text over image407
  • Network Camera with AcuSense

  • M-VPro Series NVRs

  • · Core value

    - Self-iterative algorithms to improve alarm accuracy

    - Reduce the workload and increase the efficiency of security staff

  • · Application scenarios

  • Factories

  • Industrial parks

  • Network Camera with AcuSense
  • M-VPro Series NVRs
  • · Core value

     
    - Self-iterative algorithms to improve alarm accuracy
    - Reduce the workload and increase the efficiency of security staff
  • · Application scenarios

  • Factories
  • Industrial parks

Want to learn more?

Explore other technologies

Hikvision.com uses strictly necessary cookies and related technologies to enable the website to function. With your consent, we would also like to use cookies to observe and analyse traffic levels and other metrics / show you targeted advertising / show you advertising on the basis of your location / tailor our website's content. For more information on cookie practices please refer to our cookie policy.

 

Contact Us
Hik-Partner Pro close
Hik-Partner Pro
Security Business Assistant. At Your Fingertips. Learn more
Hik-Partner Pro
Scan and download the app
Hik-Partner Pro
Hik-Partner Pro

Get a better browsing experience

You are using a web browser we don’t support. Please try one of the following options to have a better experience of our web content.