Vol 8 , Issue 2 , April - June 2020 | Pages: 48-54 | Research Paper
Received: January 17, 2020 | Revised: May 20, 2020 | Accepted: May 28, 2020 | Published Online: June 15, 2020
Author Details
( * ) denotes Corresponding author
Tracking multiple persons is a challenging task when persons move in groups and occlude each other. Existing group based methods have extensively investigated how to make group division more accurate in a tracking-by-detection framework. However, few of them quantify the group dynamics or consider the group in a dynamic view. Inspired by the sociological properties of pedestrians, this work proposes a network that tracks the moving persons.This work uses CNN algorithm by extracting the ROI based HOG features to track more accurately without any interference and to obtain a robust free result. It can be done by considering the thresholding values of the person and based on the thresholding bboxes are assigned to the persons to keep the track s of the persons being detected. Implementation of algorithm, creation of user Interface, leads to observe the performance criteria of the persons being tracked which will gives us accurate and robust free results and widely used in video surveillance and generates direct response. As in the case of any accidental decisions taken by a manual observation can be replaced by using this kind of network. CNN based tracking can overcome the problem of manual observation very accurately based on region of Interest.
Keywords
Surveillance video; Multiple person detection; Convolutional neural network; Detection accuracy; Robust detection; Occlusion