Training on Smart Functions and Applications

Hikvision Certified Security Associate

HCSA 1
Contents

 Behavior Analysis
 People Counting
 Heat Mapping
 Queue Detection
 Face Recognition

HCSA 2
Contents

 Fundamentals
 Installation
 Behavior Analysis
 Configuration
 People Counting  Case study
 Heat Mapping
 Queue Detection
 Face Recognition

HCSA 3
Background Modeling Technology
 Background modeling technology creates static background model in the
monitoring area and at the same time detects and records the moving objects. It
will trigger the alarm if the moving objects reach the alarm threshold.
Background

Line crossing
detection
RULE Region entrance
• enter detection
• leave Region exiting
• appear detection
• Color • disappear
• Shape Intrusion detection
• Size
• Type
• Path

Foreground
HCSA 4
Background Modeling Technology

 Background modeling technology is one of the most mature and widely

applied technology used in intelligent video analysis.

Advantages Challenges

High sensitivity Affected by the environment:

Various rules occasional light changes (car light, light
reflection, backlight, etc. ), leaves,
Low installation requirement
moving shadows, insects, rapid change
of weather, etc.

HCSA 5
Line Crossing Rule
 Line crossing detection is detected by the area that the object crosses the
line. By default, if the object crosses the line by 50%, it will trigger the alarm.
The higher sensitivity, the less area of the object it requires to cross the line.

Line crossing detection

100% 40% 60%

HCSA 6
Region Detections
All the region detections will detect whether the center of the target
has triggered the rule.

Region entrance
detection

Region exiting
detection
4 3 2 1

Intrusion
detection

HCSA 7
Contents

 Fundamentals
 Installation
 Behavior Analysis
 Configuration
 People Counting  Case study
 Heat Mapping
 Queue Detection
 Face Recognition

HCSA 8
Installation Recommendation
• Avoid to deploy cameras in scenes with excessive and fast changing light.
• Avoid scenes with glass, tiles, and other reflective objects.
• Avoid scenes with many winged insects.

Too Dark Too crowded

Line too close to edges Leaves

HCSA 9
Outdoor Installation

 The height of the target should be less than 1/3 of the image and
bigger than 1/15 of the image.

2015-4-15 12:14 2015-4-15 12:14

1/3

Too Big & Too small Proper size

HCSA 10
Object Pixel Requirement

• With the video resolution as CIF, the object pixel has to be between 5×5
and 100×100.
352

288 CIF 5×5 < < 100×100

Video resolution min. object pixel max. object pixel

CIF (352×288) 5×5 100×100
4CIF (704×576) 10×10 200×200

HCSA 11
Light Condition

 The camera is recommended in scenarios without strong

backlight to ensure the accuracy.

 Darkfighter is recommended in low light environment.

 Lightfighter is recommended in scenarios with strong backlight.

HCSA 12
Installation Recommendation
Line crossing detection
 Line crossing detection is recommended for outdoor scenarios, not indoor ones.
 You should draw a suitable line to make sure that the target can be easily
detected.

Line

Direction
Direction

Line

HCSA 13
Installation Recommendation
Line crossing detection
 The camera should be mounted in a comparatively high position.
 The target cannot be too big.

Target

Line

HCSA 14
Installation Recommendation

Intrusion detection
Due to the camera mounting position and direction, the target is too big in the image.
Although line crossing can be used in this scenario, but in order to avoid wrong detection,
intrusion detection is more preferred in this situation.

HCSA 15
Contents

 Fundamentals 1. Line crossing detection

 Installation 2. Intrusion detection

 Behavior Analysis 3. Region entrance detection
 Configuration
4. Region exiting detection
 People Counting
 Case study 5. Unattended baggage detection
 Heat Mapping 6. Object removal detection

 Queue Detection
 Face Recognition

HCSA 16
Line Crossing Detection
Demo Video

Direction:
A->B：Targets that moving from A to B will trigger alarm.
B->A：Targets that moving from B to A will trigger alarm.
A<->B：Targets that moving between A and B will trigger
alarm.
HCSA 17
Line Crossing Detection
Demo Video

Sensitivity:
It stands for the percentage of the body part of an
acceptable target that goes across the pre-defined line.

HCSA 18
Line Crossing Detection

 A – target area that has already crossed the line or entered the region.
 B – target area that has not crossed the line or has not entered the region
 A+B: the complete target area
 Sensitivity: B/(A+B), meaning how much the target has crossed the line or
entered the region.
HCSA 19
Intrusion Detection
Demo Video

Threshold:
The threshold for the time of the object loitering in the region. Alarm will
be triggered immediately after the object entering the region if the time
exceeds the threshold. The unit time is 1 second.

Sensitivity：
Range [1-100]. Sensitivity stands for the percentage of the body part of
an acceptable target that enters the pre-defined region.

HCSA 20
Region Entrance Detection
Demo Video

Sensitivity:
Range [1-100]. Sensitivity stands for the percentage of the body part
of an acceptable target that enters the pre-defined region.

HCSA 21
Region Exiting Detection

Demo Video

Sensitivity:
Range [1-100]. Sensitivity stands for the percentage of the body part
of an acceptable target that exits the pre-defined region.

HCSA 22
Unattended Baggage Detection
Demo Video

Threshold:
Alarm will be triggered immediately after the object being left if
length of time exceeds the threshold.
Sensitivity:
Range [1-100]. Sensitivity stands for the percentage of the body part
of an acceptable target that enters the pre-defined region.

HCSA 23
Object Removal Detection
Demo Video

Threshold:
Range [5s-20s], the threshold for the time of the objects removed
from the region.
Sensitivity:
Range [1-100]. Sensitivity stands for the percentage of the body part
of an acceptable target that leaves the pre-defined region.

HCSA 24
Arming Schedule & Linkage Actions

In order to make the smart

functions work, don’t forget to set
the arming schedule. By default, it
is armed 24/7.

Linkage methods can be set

accordingly.
• Notify Surveillance Center: push
notifications to VMS
• Upload to FTP: capture image and
upload to FTP
• Trigger Channel: trigger to record

HCSA 25
Contents
 Fundamentals
 Installation
 Behavior Analysis
 Configuration
 People Counting
 Case study
 Heat Mapping
 Queue Detection
 Face Recognition

HCSA 26
 Line Crossing or Intrusion Detection ?
Case Study: If someone go cross the fence and get into the yard, alarms should be
triggered. But there are always some people in the yard entering and leaving the protected
area which the alarm shouldn’t be triggered in this case.
Solution 1: If you draw a protected area as the following picture shows, setting the rule as
intrusion detection, the people who are already in the yard will trigger the false alarm.

Someone will
Rule move here.

Someone will
move here.

HCSA 27
 Line Crossing or Intrusion Detection ?
Solution 2: As the picture shows below, draw lines over the fence and set line crossing rules. In this way,
people inside or outside of the yard will not trigger the alarm, but only people who get over the fence will
trigger the alarm. In this case, Line Crossing rather than Intrusion Detection can improve the accuracy of the
detection.

Rule Someone will

move here.

Someone will
move here.

HCSA 28
 Line Crossing or Intrusion Detection ?
Case Study: People who get into the garage at night is prohibited. Anyone entering will trigger the alarm.
Solution 1: As is shown below, draw Line Crossing Rules all around the garage. As the lines are too close
to the image edge, camera would not be able to detect fast moving target when it enters the garage, which
will cause missing alarms.

Rule

HCSA 29
 Line Crossing or Intrusion Detection ?
Solution 2: Draw the area with intrusion detection rules. As long as moving targets show in the protected
area and stay for more than the preset time, an alarm will be triggered.
Tip: Intrusion Detection rather than Line Crossing Detection is recommended here, because Intrusion
Detection can reduce missing alarms more effectively.

Rule

HCSA 30
Contents

 Behavior Analysis  Overview

 Installation
 People Counting
 Configuration
 Heat Mapping
 Queue Detection
 Face Recognition

HCSA 31
Introduction
The People Counting is used to calculate the number of people entered or
left a certain configured area and form in daily/weekly/monthly/annual
reports for analysis.
Enter：210 Enter：110 Enter：320
Leave：190 Leave：140 Leave：90

Enter：210 Enter：330 Enter：20

Leave：190 Leave：290 Leave：10

HCSA 32
Application Scenarios

Shopping Mall Chain Stores

Supermarket
HCSA 33
Topology

HCSA 34
People Counting IPC

DS-2CD6825G0/C-I(V)(S)
Key features:
 Embedded intelligent algorithm to calculate the number of the people entering or leaving.
 Daily, weekly, monthly, annual report
 Built-in Flash for people counting statistics storage without SD card.

HCSA 35
People Counting IPC

DS-2CD6825G0/C-I(V)(S)
Features:
• Dual lens
• Support stereo views
• 2.0/2.8mm optional
• DC12V/PoE
• Specialized people counting function, support separate counting of people entering,
exiting

HCSA 36
Dual lens —3D Image

Every object in the image has X, Y and Z

axes coordinates. The Z-axis coordinate
represents height, which prevents the
interference from other objects.

Advantages:
• More accurate, exceeds 97%
• Easier configuration
• Better anti-interference ability
• Small & compact

HCSA 37
 Dual lens — advantages

• Shadow or other objects will not cause • Based on the height information in the 3D
miscounting. axis, the target height should be in the
• Camera can identify two person as long as there is range of the human height, thus filtering
space between two heads. the interference of the cart, baby carriage,
etc.

HCSA 38
Dual lens — advantages

+1

• The camera can focus on a target and track it

until it disappears in the image, thus • Provide the counting data of people
effectively filtering the repeated count of the passing by to help analyze entering rate.
loitering target.

HCSA 39
Demo Video

Enter
Leave

HCSA 40
Contents

 Behavior Analysis  Overview

 Installation
 People Counting
 Configuration
 Heat Mapping
 Queue Detection
 Face Recognition

HCSA 41
Installation recommendation
How to choose a right position?

The camera should be installed in the middle of entrance

door. Also it is recommended that the camera’s position
should be as closed as possible to target entrance door
without any obstacles.

HCSA 42
Installation recommendation

How to define the camera number?

It requires one camera for a standard entrance door of width between 2 ~ 3m and height between 2.5~5m.
* The real camera’s effective coverage varies based on camera’s focal length as the table shows.

HCSA 43
Installation Recommendation
DS-2CD6825G0/C-I(V)(S)

Focal Length (mm) Mounting Height (m) Max. Detection Span (m)

2 2.1 1.02
2 2.6 2.30
2 3.1 3.58
2 3.6 4.86
2 4.1 6.14
2 4.6 5.50
2 5.1 6.45
2 5.6 7.40
2 6 8.16

HCSA 44
 Installation Recommendation

How to define the camera number?

The number of cameras should be increase if:

- The target door is wider than camera’s effective coverage.
* camera’s effective coverage is defined by the detection line via IE client or iVMS software instead of camera’s physical FoV.

HCSA 45
 Installation Recommendation

How to define the camera number?

The number of cameras should increase if:
- There are obstacles in camera’s FoV, such as anti-theft detector, turnstile etc. which cover the most part of target people.

HCSA 46
Installation Requirements

HCSA 47
Contents

 Behavior Analysis  Overview

 People Counting  Installation

 Heat Mapping  Configuration

 Queue Detection
 Face Recognition

HCSA 48
People Counting on web browser
DS-2CD6825G0/C-I(V)(S)
 VCA > People Counting

Enable People Counting

Set Lens height

HCSA 49
People Counting on web browser
DS-2CD6825G0/C-I(V)(S)
 VCA > People Counting

 Draw a Detection Line to define a valid range;

 Arrow points to the direction of entering;
 In manual calibration, we need set the height of camera
 In auto calibration, we need draw a green square aimed at any
area of the flat floor for calibration

1 Detection line (yellow)

2 Detection Area (red)

3 Calibration Area (green)

HCSA 50
People Counting on web browser
DS-2CD6825G0/C-I(V)(S)
 VCA > People Counting

Passing by
Entering

• “Enter quantity” adds 1 if one target appears in “A”

region, crosses the yellow line, and then disappears
from “B” region. On the contrast, “Exit quantity”
adds 1. If targets just pass through A region, then
“pass by quantity” will increase.

HCSA 51
People Counting on web browser
DS-2CD6825G0/C-I(V)(S)
 VCA > People Counting > Data Uploading

Real-time Upload Data

HCSA 52
People Counting on web browser
DS-2CD6825G0/C-I(V)(S)
 VCA > People Counting > Advanced

HCSA 53
People Counting on web browser
iDS-2CD6412FWD/C &iDS-2CD6124FWD-I(Z)C
 Click Counting to export the data.
 Select to export the Statistics Result as Table, Bar Chart, or Line Chart.
Set report type, statistics type and time,
then click counting to generate report.

HCSA 54
 People Counting on iVMS-4200
• Click the report icon on the control panel to open the report page, select people counting.

HCSA 55
People Counting on iVMS-4200
 Select the report type as needed and set the time.
 Select Enter, Exit, or Enter and Exit as the data type.
 Select the camera(s) for generating the report.
 Click Search and the statistics display on the right panel. The detailed data for each hour, day or month will
be also displayed.

HCSA 56
People Counting on Local NVR
 Enter the People Counting interface: Menu>VCA Search> People Counting
 Select the camera, report type and time.
 Click the Counting button.

Set report type, statistics time, then

click counting to generate report.

HCSA 57
Contents

 Behavior Analysis
 Overview
 People Counting
 Product &Installation
 Heat Mapping
 Configuration
 Queue Detection  Statistics Report

 Face Recognition

HCSA 58
Application Scenarios

Supermarket Shopping mall

Plaza Luxury store

HCSA 59
Heat Mapping

Daily View The Heat Value

( the accumulative dwell time)
high

Weekly View

Monthly View low

HCSA 60
Heat Mapping Algorithm

Foreground
analysis

Heat accumulation

Color display

HCSA 61
Contents

 Behavior Analysis
 Overview
 People Counting
 Product &Installation
 Heat Mapping
 Configuration
 Queue Detection  Statistics Report

 Face Recognition

HCSA 62
Heat Mapping Camera

DS-2CD63X5G0-I(V)(S)
• 6MP/12MP
• 360°view angle
• Up to 15m IR range
• Built-in microphone & speaker
• Deep Learning Algorithm

HCSA 63
Installation

Objective: cover the biggest view in the area.

HCSA 64
Contents

 Behavior Analysis
 Overview
 People Counting
 Product &Installation
 Heat Mapping
 Configuration
 Queue Detection  Statistics Report

 Face Recognition

HCSA 65
Heat Mapping Operation
 Configuration > Advanced Configuration > Heat Map
 Enable heat mapping function and draw the detected area.

HCSA 66
Contents

 Behavior Analysis
 Overview
 People Counting
 Product &Installation
 Heat Mapping
 Configuration
 Queue Detection  Statistics Report

 Face Recognition

HCSA 67
Heat Mapping Operation
 Click Counting to export the data.
 Select Statistics Result as Space Heat Map or Time Heat Map, and the corresponding heat map will be displayed.
 As shown in the figure above, red color block indicates the most welcome area, and blue color block indicates the
less-popular area.

• You can get people flow and know

when is the rush hour.

HCSA 68
Heat Mapping on iVMS-4200
• Add a heat mapping network camera.
• Configured heat mapping rule: Device management-remote configuration-VCA configuration

HCSA 69
Heat Mapping on iVMS-4200
• Click to select a heat mapping camera in the area panel.
• Select the report type as needed and set the start time.
• Click Generate Heat Map and the heat map of the camera displays.

HCSA 70
Heat Mapping on Local NVR
 Enter the Heat Map interface: Menu>VCA Search> Heat Map
 Select the camera, report type and time.
 Click the Counting button to export the report data and start heat map statistics, and the results are
displayed in graphics marked in different colors.

HCSA 71
Contents

 Behavior Analysis
 People Counting
 Overview
 Heat Mapping
 Features
 Queue Detection  Configuration
 Face Recognition  Statistics Report

HCSA 72
Flow Analysis Introduction
Pain Points in Retail Industry
Pain Points in Retail Industry
Queue for a long time in supermarket! ——Poor Efficiency
Queue for a long time in supermarket! ——Poor Efficiency

Automatic queue monitoring and alarm

HCSA 73
Flow Analysis Introduction

City B

RSM

City A

Hik-Central
Hik-Central Client

Branch 01 Branch 02

NVR NVR

ID tag & waiting time Accurate counting Flow Analysis Cameras

of each person of people
Flow Analysis System
HCSA 74
Contents

 Behavior Analysis
 People Counting
 Overview
 Heat Mapping
 Features
 Queue Detection
 Configuration
 Face Recognition  Statistics Report

HCSA 75
Flow Analysis Introduction

• Waiting time of each person

Waiting time alarm!

• The number of queuing-up people

• Alarm interval prevents repeated alarms

Region \ Alarm Threshold People number Waiting time /s Alarm interval /s

Region1 15 30 10
Region2 5 5 10

HCSA 76
Flow Analysis Introduction

• Waiting time of each person

• The number of queuing-up people

People number alarm!

• Alarm interval prevents repeated alarms

Region \ Alarm Threshold People number Waiting time /s Alarm interval /s

Region1 10 30 20
Region2 5 5 20

HCSA 77
Flow Analysis Introduction

• The number of queuing-up people

• Waiting time of each person

• Alarm interval prevents repeated alarms

13 s  23 s  33 s alarm!

Region \ Alarm Threshold People number Waiting time /s Alarm interval /s

Region1 15 30 10
Region2 5 5 10
HCSA 78
Contents

 Behavior Analysis
 People Counting
 Heat Mapping  Overview

 Queue Detection  Features

 Configuration
 Face Recognition
 Statistics Report

HCSA 79
Configuration

HCSA 80
Contents

 Behavior Analysis
 People Counting
 Heat Mapping  Overview

 Queue Detection  Features

 Configuration
 Face Recognition
 Statistics Report

HCSA 81
Flow Analysis Introduction

Queuing-Up Time Analysis

Statistics Type Region Waiting Time Level Report Type Statistics Time

Regional Comparison Region1 Shorter than ____ s Daily Report

Multi-Level Comparison Region2 ____ - ____ s Weekly Report

Region3 Longer than ____ s Monthly Report

Queue Status Analysis

Queue Length Level

Fewer than ____ persons

____ - ____ persons

More than ____ persons

HCSA 82
Flow Analysis Introduction

Multi-level comparison for each region at 2 — 3 levels, for example

Waiting Time /s Color

Level 1 shorter The number of people whose waiting
Waiting time <= 5 s
than 5 s time is at level 1
The number of people whose waiting
Level 2 5 - 10 s 5 s < Waiting time <= 10 s
time is at level 2
Level 3 longer than The number of people whose waiting
Waiting time > 10 s
10 s time is at level 3

HCSA 83
 Flow Analysis Introduction

How many customers for one same queue within a period of time?

313
819 Provide the number of customers
received in the same queue within a
1481 period of time. Compare different
queues (region 1 & 2) in the same
period.

Waiting time <= 5 s 5 s < Waiting time <= 10 s Waiting time > 10 s

HCSA 84
Flow Analysis Introduction

Multi-level comparison for each region at three levels, for example

Queue Length
Color
(persons)

Level 1 fewer The percentage / period of time when Short queue: 1800 seconds
than 3 the queue length status is at level 1 within an hour

The percentage / period of time when Medium queue: 1080 seconds

Level 2 3 - 5
the queue length status is at level 2 within an hour

Level 3 more The percentage / period of time when Long queue: 720 seconds
than 5 the queue length status is at level 3 within an hour

HCSA 85
 Flow Analysis Introduction

How to improve the service strategy to deal with peak times?

Region1 Queue Status Line Chart

Queue Length <= 3 persons 3 < Queue Length <= 5 persons Queue Length > 5 persons
100%
Region1 Queue Status Line Chart
PERCENT Percent

Queue Length <= 3 persons 3 < Queue Length <= 5 persons Queue Length > 5 persons
50% Region 1_Level 1 Region 1_Level 2 Region 1_Level 3
100%
0%
80%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
60% TIME
40%
Region2 Queue Status Line Chart Compare the queue length status of the
20%
0% Queue Length <= 3 persons 3 < Queue Length <= 5 persons Queue Length > 5 persons same queue at different times.
Percent

100%
80% Compare the queue length status of
60%
40%
TIME
different queues (region 1 & 2) at the
20%
0% same time.
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
TIME

HCSA 86
Contents

 Behavior Analysis
 People Counting
 Heat Mapping
 Queue Detection  Fundamentals

 Face Recognition  Installation

 Configurations

HCSA 87
Face Recognition

Deep learning is intrinsically different from other algorithms. The way it solves the insufficiencies of
traditional algorithms is encompassed in the following aspects.

 From “Shallow” to “Deep”

 From “Artificial Features” to “Feature Learning”

Blurred Image Side Face Small Target

75.48% 81.68% 84.85%

45% 65% 195%
52.11% 49.36% 28.79%

Traditional algorithm Deep learning algorithm

HCSA 88
Face Recognition

Face Capture Face Comparison Similarity

Modeling

HCSA 89
Face Capture

Original Video

Face Detection

Face Tracking

Face Grading Deflection Sharpness Brightness

Optimal face picture

Powered by deep learning algorithm, the camera can accurately and efficiently detect moving targets in the scene.
When there is a human face in view, it will keep tracking the target and grading the detected face on each flame. The
grading is based on the definition and deflection angle of the captured face picture, and an optimal picture will be
presented accordingly.

HCSA 90
Face Detection

HCSA 91
Face Recognition

Face Capture Face Comparison Similarity

Modeling

HCSA 92
Face Recognition

Deep
Learning

Based on Deep Learning GPU

Algorithm + GPU

Facial Modeling: building face templates (n-dimensional eigenvector) by extracting

key features of human faces.

HCSA 93
Face Recognition

Face Capture Face Comparison Similarity

Modeling

HCSA 94
 Hikvision DeepinView Cameras
Face Recognition Camera (iDS-2CD7xxxG0)

Live view
(Detection) Comparison Results
(Recognition)

Real Time Captured Face Human & Crowd

(Capture) HCSA 95
Contents

 Behavior Analysis
 People Counting
 Heat Mapping
 Queue Detection  Fundamentals

 Face Recognition  Installation

 Configurations

HCSA 96
Installation Recommendation

1. The camera should be installed in the front of passageway and capture the positive face. The horizontal angle of deflection had
better be less than 25 °.
2. The vertically elevation angle should meet the below requirement: α=15±5°.
3. It requires that the covered PD (Pupil Distance) pixels needs to be larger than 40 pixels to distinguish the face details from the
captured image. The practical width of face detection should be less than or equal to 3 meters with 2MP camera.
4. Make sure that there is no shelter between the camera lens and the passageway

HCSA 97
Examples

Suitable Example Unsuitable Example

HCSA 98
Contents

 Behavior Analysis
 People Counting
 Heat Mapping
 Queue Detection  Fundamentals

 Face Recognition  Installation

 Configurations

HCSA 99
Face Recognition

HCSA 100
Face Recognition

HCSA 101
Face Recognition

HCSA 102
Face Recognition

HCSA 103
Face Recognition

HCSA 104
Face Recognition

HCSA 105
Facial Recognition _to be continued

Real-time Alarm

1V1 Comparison
Face Pictures
Search face
by face

Confirm
Identification
Video Stream
Support Smart IPC DS-IE6316/32-EL/FA
for video analysis High Frequency
Alarm

HCSA 106
Thanks!

HCSA 107