Smart Intersection

Smart Intersection

Integrated intelligent unmanned crackdown system applied with AI-Deep learning algorithm
Scope to apply
This is the specification on the Integrated intelligent unmanned crackdown system applied with AI-Deep learning algorithm of which individual parts and software are modularized to be constructed in various types for the common performance based on the basic system.
Features of Product
  • System modularization / minimization
    Innovative system integrated through modularization of multiple functions for the first time in the world
  • CPU parallel processing
    Enhancement of processing speed with embedded high speed hardware module dedicated to image process of MFSC
  • AI DEEP LEARNING ALGORITHM
    Intelligent multi-function modular camera with AI Deep learning algorithm
  • Low power dissipation
    Solar system applicable to the area without power line at a quarter of power cost
Features of product and Comparison to other ones
1. This system recognizes the information of vehicles together with their movements through the AI-Deep learning algorithm application and the two technologies patented as "Video stabilization method based on SURF for enhancing accuracy and reliability" and "Detection of vehicle video and license plate region extraction method".
Existing system to crack down
(using the loop sensors)
Smart Intersection Management System
(using vehicle detection and tracking system to crack down)
AI / Enhancement of recognition accuracy 1. Cross-correlation
2. Conventional AI Algorithm
1. Faster - CNN Al algorithm
2. YOLO AI - deeplearning
Enhancement of pre-processing capability 1. Conventional Filtering
2. Histogram normalizing
1. Image Stabilization with SURF feature point tracking / Image sensing by Blob
Crackdown of violations Crackdown of traffic signal violation using loop sensors linked to traffic signals Crackdown of violations on traffic signals, intersection traffic, U-turn and left turn with detecting and tracking of vehicles and synchronizing of traffic signals
Security Backdoor risks incurred from 'Made in China' Enhancements of security and performance with the hardware and the software algorithm that were developed in house
Others Budget waste incurred from non-integrated R&D on a single purposed project Prevention of budget waste through multi-functional combination / construction with indepfinishent modules in accordance with respective purposes
2. This system has the cameras for vehicle detection/tracking, license plate recognition, traffic signal recognition and object recognition and recognizes the vehicles in violations of the Traffic Law on traffic signals, speed limits, intersection traffic, centerline inversion(U-turn) etc. by synchronizing the information from various devices.
MFSC(Multi-Functional Smart Camera)
This system is to prevent the waste of the government budget by eliminating redundant cameras overlapped in purposes and integrating them into a single system.
Existing crackdown system Technology applied to MFSC
Modules to detect and recognize Cross-Correlation, 95% of recognition accuracy AI Deep Learning, 99% of recognition accuracy (a fifth of unrecognized rate)
Processor Industrial Computer (Averange 3GHz clock) Parallel processing Embedded Processors 20 TOPS (20 times at least)
Structures of Hardware and Software All-in-one type Modular type for variable configuration
Range to cover Class 1 Integration type of multi-functions (of 6 kinds)
Range to detect and recognize The 1st lane 2 or 3 lanes
LPR function Dedicated to LPR LPR and Image object processing
Power (for the whole system) 300W 100W (1/3 power consumption)
Size (of the processor) 430 X 430 X 90 mm 80 X 100 X 70mm (1/30 in volume)
Weight (of the processor) 2.6 kg 0.5kg (1/5 in weight)
Determinable of violation or not
  • Existing crackdown system
    Impossible to crack down on U-turn violations because of no connection between loop sensors and LPR linked to traffic signals
  • Smart Intersection Crackdown System
    Sure to crack down on U-turn violations with synchronization between traffic signals and detecting/tracking of vehicles
  • Existing crackdown system on traffic signal violations
    Impossible to crack down on left turn violations with recognizable of vehicles only in straight lanes
  • Smart Intersection Crackdown System
    Sure to crack down on both violations in left turn and red light with synchronization between traffic signals and detecting/tracking of vehicles
  • Existing crackdown system
    Impossible to crack down on violations in Intersection traffic methods because of unrecognizable of license plates of tracked vehicles synchronizing with green lights
  • Smart Intersection Crackdown System
    Sure to crack down on violations in Intersection traffic methods with synchronization between traffic signals and detecting/tracking of vehicles
System Configuration
Configuration on Integrated unmanned crackdown camera system at intersection
Example of Intersection crackdown systems to orientations of violations
  • System configuration to cover the entire intersection of max. 4 directions
  • Tracking-camera installation to cover 2 or 4 directions depfinishing on the size of an intersections
  • Operations of cameras to collect the vehicle information(MFSC, to cover 1 or 3 lanes) and recognize the traffic signals
  • 2 sets at least of MFSC on the case of more than 4 lanes
  • Tracking camera that estimates traffic volumes and averange speeds in traffic jam by analyzing the vehicle flows
  • The dotted circles on the installation figures represent the detection coverages in respective directions.
FLOW CHART ON PROGRESS OF K-INNOVATIVE PRODUCT
Products for Smart Intersection