A Novel Architecture for Object Detection and Tracking

A Novel Architecture for Object Detection and Tracking

J.Thilagavathy                                      R.Mercy

Assistant professor                               PG Student

Applied Electronics and Engineering

Grace College Of Engineering, Tuticorin

ABSTRACT

The Convolutional Neural Networks (CNN) is used for the image recognition in the field of computer vision.  Convolutional single–stage object detectors have been able to efficiently detect object of various sizes using a feature pyramid network. However, because they adopt a too simple manner of aggregating feature maps, they cannot avoid performance degradation due to information loss. The global information extractor is designed so that each feature vector that can reflect the information of the entire image through a proposed deep neural network with performance enhancement. Feature fusionmodel along with SE each channel   so as to extract all necessary features support to improve the accuracy of the detector. Object detection and tracking is one of the most common and demanding tasks that system need to perform in order to detect meaningful events and activity to automatically comment and retrieve video content. The reason object detection and tracking are grouped is that object detection can be considered the basis of object tracking, and everyone needs to choose the right features and train for effective classification.