Impact of Feature Representation Quality on Person Re-Identification Accuracy Using Various Extraction Methods
P. Merlin, Paul T Sheeba · 2025
Person Reidentification is a key technology in the modern security system due to the growing public security. It can be elucidated as retrieving a person's image from the enormous number of images obtained by the surveillance cameras at different locations. The key to re-identifying the individuals is to learn the distinguishing traits of the precise person image, as these feature descriptors aid ReID systems in differentiating between people. Earlier, Handcrafted algorithms were deployed and Deep Learning has now transformed feature learning using neural networks. Hybrid methods are also proposed that merge both features. This paper discusses the various approaches for learning the distinguishing characteristics and highlights the accuracy derived from each method. The experimental results have proved that Deep feature method yields good accuracy.