Minoh Laboratory > Datasets > Shinpuhkan dataset(s)

Shinpuhkan dataset(s)


How to get the dataset

Please download, read, and sign this agreement.

After you send it to us ( shinpuhkan_datasets (at) mm.media.kyoto-u.ac.jp ),
we will tell you the URL of the dataset download page.

Agreement

Version 2014

Abstract

We present a novel manually annotated dataset which contains multiple tracklets for each person captured by multiple overlapping/non-overlapping cameras in surveillance scenarios.

This dataset consists of more than 22,000 images of 24 people which are captured by 16 cameras installed in a shopping mall "Shinpuh-kan".
All images are manually cropped and resized to 48x128 pixels, grouped into tracklets and added annotation.
The number of tracklets of each person is 86.
This dataset contains multiple tracklets in different directions for each person within a camera.

This dataset is named "Shinpuhkan2014 dataset".

All subjects

Shinpuhkan2014 dataset includes the following subjects.

Reference

If you evaluate performance of your methods includes re-identification, image retrieval and recognition, please refer to the following paper: