Residual Network based Single Object Tracking

In this paper, we propose the idea of single object tracking using RESNET-101. The proposed Residual Network based Single Object Tracking (R-SOT) consists of the basic idea of supervised learning. The frames are given as an input to the tracker and the tracker predict the bounding boxes and place them on a specified object and move that bounding box with that object in the entire frames of video. The detection of bounding boxes from RESNET-101 is performed by using Region Based Convolutional Neural network (RCNN) Object Detector. The performance of the proposed tracker is checked using Online Object Tracking Benchmark(OOTB) dataset. Test Results show that the proposed tracker is able to track the single object with high precision.