Volume 14 Issue 3 - September 2019

  • 1. Surveillance video analysis using deep learning techniques for traffic and crowd management

    Authors : S. Seema, Suhas Goutham, Smaranita Vasudev, Rakshith R Putane

    Pages : 1-6

    DOI : http://dx.doi.org/10.21172/1.143.01

    Keywords : Deep learningSingle Shot MultiBox DetectorLine of counting approachOptical Character RecognitionTesseractSurveillance video analysis

    Abstract :

    In this digital age, the availability of massive amounts of video data have been exploited by deep learning techniques to gain useful insights. The efficiency and processing power of terabytes of video data using deep learning techniques has been extensively researched on and been used in the field of object counting. Applying deep learning algorithms on surveillance video data can help in the areas of traffic and crowd management. The model proposed for achieving these objectives is Single Shot MultiBox Detector (SSD) with a line of counting approach to count the objects of interest from a surveillance video. The proposed model has been used for analyzing traffic surveillance videos for counting of vehicles on different lanes and make intelligent traffic decisions to prioritize traffic signals based on the traffic densities over a period of time. As a subcase of traffic management, a Tesseract OCR model is run on surveillance videos to capture the license plate of vehicles violating any traffic regulations. Another use case of object counting proposed in this paper involves studying and analyzing the crowd statistics from publicly accessible surveillance video cameras, to handle crowd management in cases of emergencies and huge public gatherings for safety and security. The need for accuracy along with robustness makes deep learning a suitable choice for the use cases enlisted.

    Citing this Journal Article :

    S. Seema, Suhas Goutham, Smaranita Vasudev, Rakshith R Putane, "Surveillance video analysis using deep learning techniques for traffic and crowd management", Volume 14 Issue 3 - September 2019, 1-6