A Comparative Analysis of Visual Object Tracking Approaches & Impact on Benchmarks

Visual Tracking is an approach through which a moving object can be tracked at distinct pixel points and coordinates. It is a part of machine vision technology where a machine can recognize the objects as per the patterns and track in the whole frame accordingly. There are so many challenges to recognize the object and track without any overflow. Conventional model is not able to detect the object in the entire frame with high level of accuracy. Machine learning is in trending phase where a dataset can be trained to obtain the better result in tracking the object with more preciseness. There are several researches have been done in this field but certain flaws present in the existing systems for few challenges. Each and every system has its own flaws where accuracy may degrade and system encounters more overlap. There are various benchmarks are available to test the system and obtain the result accordingly to pertain the precision rate. The intention of this paper is to find out the flaws present in the existing algorithms and their results as per the benchmark selected. There are various challenges present in the benchmarks which need to be revised with better accuracy.