About this document
Deep Spatio-Temporal Encoding for Re-ID by Salma Ksibi Mejdoub is a document available to read on EtoBox.
This paper introduces a deep spatio-temporal appearance (DSTA) descriptor for person re-identification (re-ID) that utilizes deep Fisher vector encoding to robustly handle misalignment in pedestrian tracklets. The proposed method integrates saliency and Gaussian templates to enhance encoding, achieving competitive accuracy with state-of-the-art techniques without the need for pre-training or data augmentation. Experimental results on four challenging datasets demonstrate the effectiveness of the DSTA descri
- Author
- Salma Ksibi Mejdoub
- Language
- EN