System

Main processor

2 Intel Xeon Silver 4114T

Operating system

CentOS Linux release 7.4.1708 (Core)

Intelligent card

1 Dahua DH-AIX3000 self-researched standard half-height intelligent card

Memory

8 16GB DDR4 memory, maximum 24 slots.

Disk

4 3.5” 4T disks which can be extended to maximum 32T (each disk is 4T), and maximum 8 slots.
7.2K RPM SATA 6Gbps 512n 3.5”

Structuring Intelligent Analysis

Target classification

Classify targets, including humans, motor vehicles and non-motor vehicles. Make intelligent structured analysis on the live video, historical video (supported by platform) and the offline uploaded video files, and then abstract structured information about the moving targets

Motor vehicle detection

Detect vehicles by vehicle type, color, brand, model, year, plate number, and plate type. Detect sunshield, safety belt, phone call and inside decorations (pendant and tissue box).

Non-motor vehicle detection

Detect non-motor vehicles by driver's gender, age, hair style, wearing (hat), belongings (umbrella, backpack and handbag), top clothes, top color, and non-motor vehicle type.

Pedestrian detection

Detect pedestrians by gender, age, hair style, wearing (hat), belongings (umbrella, backpack and handbag), top clothes, top color, bottom clothes, and bottom color.

Region detection

Support detection area and exclusion area settings.

Real-time display

Display the analysis results in real time, and display humans, motor vehicles and non-motor vehicles in real time with tracking boxes.

Search by attribute

Support intelligent search of humans, motor vehicles and non-motor vehicles according to their attributes.

Search by picture

Select one picture manually, filter by time and channel, compare with historical data, thus find the qualified humans, motor vehicles and non-motor vehicles, and arrange them according to the similarity.

Target comparison

Compare 1V1 target quickly, and return target similarity in two pictures.

Video recording task

Make intelligent structured analysis on the historical video (supported by platform), and then abstract structured information about the moving targets. And then, search the analysis results by pictures and attributes.

Local task

Make intelligent structured analysis on the offline uploaded local video files, and abstract structured information about the moving targets. And then, search the analysis results by pictures and attributes.

Deploy attributes

Platforms

Structuring picture storage

Local storage or external cloud storage.

Display tracks on the map

Platforms

Computing node cluster

Support cluster and management of 2-XX structured analysis servers.

Structuring Intelligent Analysis Attribute Dictionary

Pedestrians

Age: Child, adult, senior citizen
Backpack: Yes, no
Umbrella: Yes, no
Handbag: Yes, no
Top color: Black, white, gray, red, green, blue, yellow, orange, purple, pink and brown. 11 colors in total.
Bottom color: Black, white, gray, red, green, blue, yellow, orange, purple, pink and brown. 11 colors in total.
Top style: Short sleeves, long sleeves
Bottom style: Shorts, pants
Gender: Male, female
Hat: No hat, hat, helmet
Hair style: Short hair, long hair, ponytail, updo, head covered.

Non-motorized vehicles

Age: Child, adult, senior citizen
Backpack: Yes, no
Umbrella: Yes, no
Handbag: Yes, no
Top color: Black, white, gray, red, green, blue, yellow, orange, purple, pink and brown. 11 colors in total.
Top style: Short sleeves, long sleeves
Gender: Male, female
Hat: No hat, hat, helmet
Hair style: Short hair, long hair, ponytail, updo, head covered.
Non-motor vehicle type: Two-wheeled vehicle, tricycle

Motor vehicles

Vehicle color: Gray, black, white, silver gray, blue, green, purple, red, orange, pink, brown and yellow. 12 colors in total.
At least 86 logos and brands in the front can be recognized.

At least 55 logos and brands in the rear can be recognized.

Vehicle type: Passenger vehicle, heavy truck, medium truck, car, van, light truck, medium bus, SUV, MPV, bus, pickup and minicar
Attributes inside the window: Pendant, tissue box, sunshield, telephone call, and safety belt.
Number plate color: Blue, yellow, white, green and black.
Recognize the number plates of 47 countries in the world.

Structuring Analysis Performance

Structuring analysis performance

Each intelligent card can analyze moving targets in 20-channel 1080p live video or 40 pictures/second. One device supports maximum 6 intelligent cards.

Search by picture performance

Compare 10 million pictures in the passenger database quickly. During hot search, operator time is 4 million pictures/second, and total search time is no more than 5 s.

Region detection quantity

Support 1 detection region in one image, and at least 10 exclusion regions in one detection region.

Moving target detection rate

With normal light in the day and normal fill light at night, resolution ≥40 × 80, clear, distinguishable and no blocking:
Detection rate of pedestrians: ≥95%;
Detection rate of non-motor vehicles: ≥95%;
With normal light in the day and normal fill light at night, resolution ≥120 × 120, clear, distinguishable and no blocking:
Detection rate of motor vehicles: ≥95%.

Recognition accuracy of pedestrians

With normal light in the day and normal fill light at night, resolution ≥80×160, clear and distinguishable pedestrians:
Gender accuracy: ≥90%;
Hair style accuracy: ≥85%;
Clothes style and color accuracy: ≥85%;
Wearing accuracy: ≥80%;
Belongings accuracy: ≥80%.

Recognition accuracy of non-motor vehicles

With normal light in the day and normal fill light at night, resolution ≥80×160, clear and distinguishable non-motor vehicles:
Gender accuracy: ≥90%;
Hair style accuracy: ≥90%;
Clothes style and color accuracy: ≥90%;
Wearing accuracy: ≥80%;
Belongings accuracy: ≥70%;
Non-motor vehicle type accuracy: ≥90%.

Recognition accuracy of motor vehicles

With normal light in the day and normal fill light at night, resolution ≥240×240, clear, distinguishable, no blocking motor vehicles whose inclined angle is no more than 25°:
Motor vehicle type accuracy: ≥92%;
Motor vehicle color accuracy: ≥90%;
Motor vehicle brand accuracy: ≥85%;
Driver and passengers’ safety belt accuracy: ≥95%;
Driver and passengers’ phone call accuracy: ≥90%;
With normal light in the day and normal fill light at night, resolution width ≥120, clear, distinguishable number plate:
Number plate accuracy: ≥96%;
Number plate color accuracy: ≥90%.

Port

Network port

Two 10000/1000Mb self-adaptive network ports

USB port

2 front USB3.0 ports and 3 rear USB3.0 ports

VGA port

2 VGA ports

Other ports

1 RJ-45 management network port

General parameter

Power supply mode

100–127V/200–240V, 50/60Hz, 10A/5A

Power redundancy

Dual power redundancy

Power consumption

≤ 800W

Operating temperature

10°C to 35°C (50°F to 95°F)

Operating humidity

35%–80% RH, maximum relative humidity during operation is 90% RH (40°C).

Storage temperature

-40°C to 60°C (-40°F to 140°F)

Storage humidity

20%–93%RH

Gross weight

35 kg (77.2 lb)

Net weight

27.5 kg (60.6 lb)

Dimensions

87.00 mm × 447.60 mm × 735.60 mm

(3.43'' × 17.62'' × 28.96'')

Package dimensions

273.00 mm × 754.00 mm × 1069.00 mm

(10.75'' × 29.68'' × 42.09'')

Mounting

Standard 19'' rack installation with guide rail.