Page 28 - JM_HikVision
P. 28

Deep Learning Technology
Deep Learning technology algorithms are the latest innovation in video image analytic systems. Hikvision leads the way in implementing these algorithms in its new DeepinView Camera series and DeepinMind NVR series of products.
For years, video content analysis software has been promoted as a way to curb false alarms and improve detection and prevention in security applications. To date, most applications have lacked accuracy and reliability. Today’s high-speed GPUs dramatically improve the processing of video data and Deep Learning is recognized as the future in technology. Hikvision is leading the way with Deep Learning products that provide users with the most reliable and accurate video content analysis.
Vertical Markets
Applications for this technology include many vertical markets that rely on more accurate and effective solutions to queue and count people, implement crowd control, manage loitering, entry/exit, vehicle management, traf c, parking and license plate recognition. Hikvision is making solutions for all of these scenarios and more with enhanced Deep Learning technology.
HikCentral/iVMS-4200
Taking advantage of this new technology does not mean replacement of existing equipment and systems. Hikvision provides DeepInView cameras that can enhance existing systems. Our HikCentral and iVMS-4200 control systems are able to manage these products. You can add a DeepInMind NVR to your system using the existing cameras on site.
R&D
The added value of Deep Learning products is that they will continue to improve as data is collected and processed through Hikvision’s Research and Development teams. Improved accuracy and reliability will be introduced with  rmware and software upgrades that take advantage of the newly “learned” subjects and characteristics.
01
01. HUMAN DETECTION
False alarms from non-human subjects (animals, vehicles, moving trees, changing lighting) are a signi cant limitation of current video analytics. Hikvision solves this problem with algorithms in cameras and NVRs
that will isolate detection of human subjects signi cantly improving video analysis accuracy and reliability.
02. FACE DETECTION
AND RECOGNITION
Facial analytics products have been in use for years with limited accuracy. Today Deep Learning algorithms can increase accuracy up to 97 percent. New methods of facial analysis, massive data collection and testing are being implemented in Hikvision products to ensure accurate face capture. High speed search and recognition analytics will provide immediate identi cation of subjects.
03. BEHAVIORAL ANALYSIS Human and vehicle characteristics
are more de ned, increasing the intelligence of these systems. Hikvision is identifying subject’s movement and positioning to help customers prevent crime before it happens. This analytic technology can help identify gender, clothing, such as hats, bags and glasses, as well as car color, shape
and model.
02
03
27
Professional Solutions


































































































   26   27   28   29   30