Surveillance systems, Person recognition, Social media data, Multi-style image generation, StyleGAN, Accuracy, Robustness
Abstract
Surveillance systems play a crucial role in maintaining security and safety in various environments. However, traditional methods of person recognition often face challenges such as poor lighting conditions, occlusions, and changes in appearance. In this paper, we propose a novel approach to enhance person recognition in surveillance systems by leveraging social media data and multi-style image generation techniques. Our system collects images of individuals from social media platforms, extracting location information if available. These images are then utilized to train a StyleGAN (Generative Adversarial Network) model, capable of generating diverse styles of a person's appearance. Subsequently, the generated images are utilized to improve the performance of person recognition algorithms in surveillance footage, thereby enhancing the accuracy and robustness of the system. We conduct experiments on real-world surveillance datasets to evaluate the effectiveness of our approach, demonstrating significant improvements in person recognition accuracy compared to traditional methods.
Article Details
Unique Paper ID: 163794
Publication Volume & Issue: Volume 10, Issue 11
Page(s): 2267 - 2274
Article Preview & Download
Share This Article
Join our RMS
Conference Alert
NCSEM 2024
National Conference on Sustainable Engineering and Management - 2024