Logistics Park Baoan District, Shenzhen City 2021-12 58

License plate recognition and freight dispatch monitoring project in a logistics park

A dual-lane entrance and exit license plate recognition and gate linkage system has been built, supporting full-scale road monitoring and platform dispatching video in the park, allowing more than 800 vehicles to pass without stopping every day.

  1. front page
  2. Project cases
  3. License plate recognition and freight dispatch monitoring project in a logistics park

Project Overview

Project basic information

Project customersA supply chain management company in Shenzhen
Project locationBaoan District, Shenzhen City
Project scale4 entry and 4 exit lanes/park road 2.2 km
Project duration50 days
Delivery time2021-12
ClassificationLogistics Park

project site

Real scene of project site

Project background

On average, more than 800 trucks enter and leave the logistics park every day. The original entrance and exit used manual registration, and the queue length exceeded 300 meters during peak hours. It not only affects the operating efficiency of the park, but also causes congestion on surrounding roads. In October 2021, the customer started the intelligent transformation of the entrance and exit.

Technical difficulties and solutions

DifficultiesSolutions
The truck license plate is seriously stainedUsing a large target area sensor + a dedicated algorithm for the license plate, and adding a stroboscopic fill light, the recognition rate of the stained license plate can still reach over 95%
Trailer multi-plate interferenceLock the main license plate through video multi-frame comparison to avoid misidentification of the tractor's front license plate
Strong light interference at nightAdopt anti-glare lens and partitioned fill light to avoid glare to the driver
Connect with the dispatching systemProvide standard API to push license plates, time, lanes and snapshots in real time

Project results

  • The average vehicle travel time is shortened from 45 seconds to 5 seconds, and the peak queue length is reduced to less than 50 meters.
  • Platform scheduling is optimized based on real-time entry and exit data, and the average waiting time for trucks has dropped by about 40%.
  • Loading and unloading cargo damage disputes are quickly identified through platform video, and the average processing time is shortened from 3 days to half a day.

Project difficulties

Issues that need to be addressed during project advancement

With an average of more than 800 trains per day, manual registration efficiency is extremely low

During peak hours, the queue of vehicles exceeds 300 meters, affecting surrounding roads.

The truck license plate is stained and the trailer has multiple license plates. The situation is complicated

Ordinary recognition algorithms have a low recognition rate for stained license plates and trailers.

The roads in the park are narrow, and crossing and reversing accidents occur frequently

Lack of road monitoring and blind spot reminders

Platform loading and unloading disputes and cargo damage are difficult to define

lack of video evidence of the loading and unloading process

Vehicle entry and exit data need to be connected with the dispatch system

Manual data entry is lagging behind and error-prone

solution

Technical and construction measures taken to address the above difficulties

Special license plate cameras are deployed in a total of 8 lanes with 4 entrances and 4 exits

including anti-glare lens and stroboscopic fill light, with a recognition rate of 99.2%

The license plate recognition is linked to the barrier, and monthly rental cars are automatically released

the pole is automatically raised after scanning the QR code to pay, and the average passage is 5 seconds

18 medium-speed infrared domes are deployed on the roads in the park

Support automatic cruise and road blind spot coverage

24 high-definition gun machines are deployed in the platform area

The entire loading and unloading process is videotaped, and cargo damage disputes can be traced

Data interface docking with customer dispatching system

Real-time transmission of vehicle entry and exit times for platform queuing dispatch

Project results

Actual operating data after the system goes online

99.2%
License plate recognition accuracy
5 seconds
average vehicle travel time
50 meters
Peak queue length (original 300 meters)
100%
Video evidence of loading and unloading process

Main equipment list

This project mainly uses equipment (XLD is mainly self-developed)

  • Special license plate cameras16 units, 8 lanes bidirectional
  • Gate all-in-one machine8 units, including anti-smash radar
  • Infrared medium-speed ball machine18 units, park road cruising
  • HD network gun machine24 units, platform loading area
  • Strobe fill light16 sets, guaranteed for night recognition
  • Industrial ring network switches8 sets, self-healing ring network