BITS-IITMandi-ForeheadCreases Images Database Version 1.0

Our work is one of the first attempts to identify humans based on their forehead regions. The motivation for this project came from the current COVID-19 situation, where all the existing biometric modalities have seen some severe drawbacks. For example, fingerprint-unlock can not always be used due to hygienic concerns and due to the usage of gloves. Face-unlock does not work when the person wears a mask, which is required in our current situation. Iris-unlock is also not suitable in phones because it requires more sophisticated hardware components, and the quality of images from smartphones is inferior to authenticate users properly. Further, iris-unlock may not work correctly when the subject is wearing spectacles. However, our work on authenticating users using forehead images works perfectly in the masked face scenarios and in every other case where the existing biometric modality fails, thus serving as a suitable alternative in COVID-19 like situations.

Description

This database has been acquired from the student volunteers and their family members of BITS Pilani, Pilani Campus, India. Our work releases the first dataset of forehead photos. The data was acquired through an Android application remotely during March-April 2021. This is acquired in an entirely unconstrained environment with changing background, lighting, and camera sensors. This dataset consists of 4,964 forehead-images from 247 subjects, with each user giving roughly 20 samples. This is the first and the largest dataset on forehead photos, which is being made public to the research community. The data was acquired in two sessions, with each subject in this dataset providing ten photos per session, with a minimum time-gap of one day between each session. We have also considered two poses, one being far from the camera (pose one) and second being closer to the camera (pose two). The subjects have provided five photos of pose one and five photos of pose two in each session. We have sufficient age and gender variation present in the dataset as all the family members of the students were also included in the database.

Sample Images

The entire database, as detailed above, is made available for the research. Our work [1], that introduces this database, utilized 'first' 247 subjects' forehead images as per the stated protocol, and these can be used for performance comparison. We provide six sample images from our dataset below.

1
2
3
4
5
6

Download

Users of this database are requested to cite the reference [1] in their work. Interested researchers should complete and submit the following form to acquire "BITS-IITMandi-ForeheadCreases Images Database Version 1.0". You will receive an email acknowledgement once you submit the form successfully. After processing the responses received, we will send further instructions on downloading the dataset in the email ID provided in the form. Commercial use/distribution of this database is strictly prohibited. The registered users of this database should seek explicit approval for publishing or posting any image(s) from this dataset.

Click Here to Access the Database

Reference

[1] Bharadwaj, R., Jaswal, G., Tiwari, K., & Nigam, A. (2022). Mobile based human identification using forehead creases: application and assessment under COVID-19 masked face scenarios. 2022 IEEE Winter Conference on Applications of Computer Vision (WACV), accepted for publication.