1. Person Re-Identification by Siamese Network
- Author
-
Newlin Shebiah Russel, S Arivazhagan, S G Amrith, and S Adarsh
- Subjects
Siamese Network , Re-idenfication , retrieval ,Electronic computers. Computer science ,QA75.5-76.95 - Abstract
Re-Identification of person aims at retrieval of person across multiple non overlapping camera. There was a huge gain in the computer vision community with the advancement of deep learning features and also the number of surveillance in videos increased. The challenges faced by person re-identification is low resolution images, pose variation etc., and convolutional neural networks are supported by a number of state-of-the-art algorithms for person re-identification. In this paper, Siamese network is used to predict the similarity or dissimilarity of a person across two cameras. It's a neural architecture that takes as input a pair of images or videos and the output as the prediction of similar and dissimilar persons along with their prediction scores. The experimentation is done by using datasets iLIDS-VID, PRID 2011 and obtained a recognition accuracy of 79.52% and 85.82% respectively.
- Published
- 2023
- Full Text
- View/download PDF