Towards Palmprint Verification On Smartphones

Yingyi Zhang1,2, Lin Zhang1,*, Ruixin Zhang2, Shaoxin Li2, Jilin Li2, and Feiyue Huang2

1School of Software Engineering, Tongji University

2YoutuLab Tencent


Introduction

This is the website for our paper "Towards Palmprint Verification On Smartphones", under review.


Tongji Mobile Palmprint Dataset

MPD 2.0(Code: ofaj)

MPD contains original palm image set and generated datasets. The sturcture of our dataset is like below:

Dataset Name

Description

Number of samples

File path

PalmSet

Original palm image set.

16000

MPDv2/original palms/

MPDD

Generated dataset for keypoint detection, in VOC format.

366816

MPDv2/generation/1-keypoint_detection/

ROI

Generated dataset of palmprint verification.

16000

MPDv2/generation/2-ROI_verification/

PalmSet: The images in original palm dataset were palm images from built-in camera of mobile phones (Huawei/Xiaomo). The palm images are collected by mobile phones in 2 periods with various background and lighting conditions. This dataset contains 16000 palmprint images from 200 sessions (400 different hands). 10 photos of one’s hand with each phones and each time period are preserved in this dataset. All palm images are labelled. Typical image samples contained in this dataset is shown below.

ROI: for a given palm image, a palmprint ROI image will be generated. The example ROI samples are shown below.

We have also developed a matlab labeling tool for labeling keypoints on palm images, whose interface is shown in the image below.

This labeling tool and the user instructions can be found here https://github.com/Shelro/MarkToolForPalmprintPoint.


Source Codes

Note: all these codes are implemented in Matlab and run on mac OSX.

1. source code.zip(Code: qc9c)

These codes can fulfill two tasks, generating annotations for keypoint detection and generating palmprint ROIs.


Last update: Feb. 24, 2020