Loading...

Automatic Traffic Incident Detection Based on Monitoring Camera Image Processing System

Mehrizi Rahmat Abadi, Ali | 2021

349 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 54596 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Gholampoor, Iman; Movahedian, Hamid
  7. Abstract:
  8. Automatic traffic incident detection is one of the most important areas in intelligent transportation systems. In recent years big cities have faced major problems with traffic flow and congestion. Therefore, real time traffic monitoring in order to analyze traffic flow plays a key role in solving traffic problems. The purpose of this study is to design an integrated system for automatic traffic incident detection and traffic monitoring based on deep learning with vehicle detection and classification, counting, average speed estimation, extracting traffic parameters such as congestion level and event detection such as unauthorized stops and deviation from the path. In this research, a dataset containing about 12100 images in 4 classes including car, truck, bus and pickup has been collected. Compared to existing datasets, our dataset is much larger and has more challenging scenes. Also, unlike existing datasets that collect images from non-traffic cameras, our dataset is collected from real traffic cameras. Another purpose of this research is to increase processing speed of designed system with appropriate accuracy. Our system is capable of working with the lowest possible resolution of 240x320 in order to optimize system bandwidth
  9. Keywords:
  10. Intelligent Transportation System (ITS) ; Surveillance Cameras ; Deep Learning ; Convolutional Neural Network ; Image Processing ; Automatic Incident Detection ; Traffic Data

 Digital Object List

 Bookmark

No TOC