A team inspired by technology with more than 400-man months of software development expertise. Qubeslab is where dream passion and work merging.

Life @ Qubeslab

Contact

Qubeslab Technologies Private Limited. 7G, Trans Asia Cyber Park, Infopark Phase II, Kakkanad 682037, Kerala, India.

info@qubeslab.com

+91-938-340-1029

Published:
March 11, 2018
Category:
Ideas / Technology
Client Location:
India
Client

Our client is a US-based technology startup serving the transportation industry with advanced AI camera solutions.

Industry

Transportation, Automation.

Overview

Automated vehicle counter solution helps the government authorities to count and classify the active vehicles in the city. This could be the most valid input for the traffic monitoring and management system.

In the recent trends of smart city projects automation is the key point. To implement effective traffic monitoring and management real-time vehicle information on each location inside the city is a must. This solution is ready to plug with the existing smart city monitoring solutions using the standard prototypes.

The system is capable to do,

  • Detect vehicles from the video camera stream.
  • Classify vehicles from an image like cars, bikes, trucks, etc.
  • Find vehicle movement direction from camera stream.
  • Capable to handle hundreds of camera streams at the same time.
  • Centralized reporting module, which will auto-generate report as per the configuration.
Qubeslab solution

To achieve the above requirements Qubeslab proposed and implemented an artificial intelligence-based video stream analyzing system, which is capable to analyze hundreds of video streams at the same time. Major modules of the system are,

  • The central video processing module
  • Video stream manager
  • The automated report generation unit
  • Web based dashboard for user

Central video processing module using the neural network image classifier which is trained with three lakh images of vehicles from different segments. The system will also do auto training using the new data to improve the accuracy of prediction. All these operations can be fully controllable by the system admin through the dashboard. The system includes multiple graphical processing units to satisfy the heavy processing power requirement.

Automated reporting module will generate daily reports including details like the number of vehicles traveling through each junction, type of vehicles, average speed, etc.

Overall, the system helped the user to avoid huge manual work. Also, it helped the authorities in the effective planning and management of heavy city traffic.

Technologies Used:
  • Qt/Qml
  • OpenGL
  • OpenCV
  • NodeJS
  • Caffe based deep learning