Analysis of Pose Estimation Based GLOGT Feature Extraction for Person Re-Identification in Surveillance Area Network

The person re-identification is the process of identifying a person of interest from the crowded scenes taken from different camera networks. With the performance saturation under different camera views and environmental settings, the research focus for person Re-ID has facing more challenging issues. Some of them are illumination, pose variation, viewpoint changes and, occlusions. To overcome these issues, we proposed a novel feature extraction method called GLOGT and pose learning-based re-identification procedure in our previous research papers. Later we came to know that the image-based analysis is more important to prove the efficiency of a novel person re-identification method. So here we conducted some important experiments to analyze the efficiency of the proposed techniques using the benchmark datasets. From the result analysis, it shows that the proposed techniques outperforming other existing techniques with a good accuracy level. Since the pose estimation-based method extracting the features based on the pose priority, reduces the training testing comparisons also. Since the GLOGT feature is a combination of three types of feature representation, one feature suppresses due to some issues, at that point the remaining will dominate and gives higher accuracy for identification.