Investigation of Downsampling Technique for Person Re-Identification Using Low Resolution Images

DOI

Bibliographic Information

Other Title
  • 低解像度に劣化した画像を用いた人物対応付けにおけるダウンサンプリング処理の効果検証

Description

We investigate whether a downsampling technique for low-resolution person images can improve the accuracy of person matching. It is general to apply a learning-based technique to extract features that are not affected by low-resolution. However, it requires a large number of pairs of high-resolution images and low-resolution images acquired from the same person. In this paper, we consider a situation that the pair of images cannot be collected. We aim to improve the accuracy of matching even in low-resolution images using a downsampling technique as a pre-processing for feature extraction. We constructed a low-resolution dataset by changing the focal length of the camera lens and the distance from the person to the camera. We confirmed that a downsampling technique improved the accuracy of the person matching compared with an upsampling technique.

Journal

Keywords

Details 詳細情報について

Report a problem

Back to top