Samarth Bharadwaj, PhD Scholar
Image Analysis & Biometrics Lab
IIIT-Delhi, India
samarthb@iiitd.ac.in
Advisers: Mayank Vatsa & Richa Singh

View Samarth Bharadwaj's profile on LinkedIn
Google Scholar
Recent CV

I am a PhD scholar in Image Analysis and Biometrics Lab at IIIT-Delhi, India. My research interests are in the area of computer vision and machine learning with applications in face biometrics. I am currently supported by the Department of Information Technology (DIT), India.

I was a visiting scholar at West Virginia University with Prof. Afzel Noore and Prof. Arun Ross in the summers of 2010 and 2011. I completed B.Tech from JIIT, Noida.


Research Topics and Select Publications

Face Spoofing Detection

Face biometric systems are vulnerable to print and replay spoofing attacks that may be detrimental to their reliability. In this ongoing study, a novel approach to spoof detection is proposed based on motion enhancement and estimation. Initial results show state-of-the-art performance in spoofing detection on two publicly available datasets.

Computationally Efficient Face Spoofing Detection with Motion Magnification, CVPR Biometrics Workshops (CVPRW), 2013.
Face Anti-spoofing via Motion Magnification and Multifeature Videolet Aggregation

 

Biometric Quality Assessment

Biometric quality indicates the usability of a biometric to uniquely identify a person. Understanding the quality of an input sample may be critical to the proper functioning of a biometric system. We find holistic orientation representations of face images to be able to encode biometric usability of a sample. This work has several practical applications in large scale biometrics. We show the advantage of quality assessment in multi-modal biometrics that are based on quality assessment of each modality. Incorporating quality in the multi-modal biometric system improves performance in adverse military conditions. The work is supported by DoD, USA.

Biometric Quality: A Review of Fingerprint, Iris, and Face, EURASIP Journal on Image and Video Processing, 2014.
Can Holistic Representations be used for Face Biometric Quality Assessment?, IEEE International Conference on Image Processing (ICIP), 2013.
Quality Assessment based Denoising to Improve Face Recognition Performance, CVPR Biometrics Workshops (CVPRW), 2011.
Quality Driven Biometric Classifier Selection Framework for Improved Performance, International Joint Conference on Biometrics (IJCB), 2011.
Texture Recognition in Ocular Biometrics

For real time on-the-move and at-a-distance personal identification,iris and face recognition must be supplemented with texture based Periocular region recognition. We propose a recognition technique that is resilient over large capture distances of up to 8 meters. This work is supported by DIT, India.

Periocular Biometrics: When Iris Recognition Fails, International Conference on Biometrics: Theory, Applications and Systems (BTAS), 2010.
Newborn Face Recognition

A robust face recognition system for newborns will help stem the high number of swapping and abduction cases, prevalent in many parts of the world, including India. It is faster and cheaper than other alternatives such as DNA testing and is more reliable than RFID bracelets.

Face Recognition for Newborns: A Preliminary Study, International Conference on Biometrics: Theory, Applications and Systems (BTAS), 2010.
Face Recognition via Derived Social Context

Humans are very efficient at recognizing familiar face images even in challenging conditions. One reason for such capabilities is the ability to understand social context between individuals. Sometimes the identity of the person in a photo can be inferred based on the identity of other persons in the same photo, when some social context between them is known. This research presents an algorithm to utilize co-occurrence of individuals as the social context to improve face recognition.

Aiding Face Recognition via Social Context Association, IEEE/IAPR Joint Conference on Biometrics (IJCB) (BTAS), 2014.
Analysis of Fingerprints of Indian Masses for Universal Identification

The feasibility of using fingerprints as a biometric modality despite challenges such as capture inaccuracies etc. were analyzed. The studies lead to quality standards of India's UID project. These results are supported by Unique Identification Authority of India (UIDAI), Government of India.

Analyzing Fingerprint of Indian Population using Image Quality: A UIDAI Case Study, In Proceedings of Workshop on Emerging Trends and Challenges in Hand based Biometrics, International Conference on Pattern Recognition, 2010.

Presentations
  • Learning based Feature Extraction Techniques for Object Recognition

    IIIT-Delhi Ketchup Talks

     

  • Algorithm Independent Pattern Recognition and Machine Learning

    IIIT-Delhi

     

  • Computationally Efficient Face Spoofing Detection with Motion Magnification

    CVPR Biometrics Workshop, 2013

     

  • Can Holistic Representations be used for Face Biometric Quality Assessment?

    ICIP, 2013

     

  • Evaluation Methodologies for Biometric Classifiers

    IIIT-Gwalior


Full Publications List
Book Chapter
  • Plastic Surgery and Face Recognition

    H.S. Bhatt, S. Bharadwaj, R. Singh, and M. Vatsa

    Encyclopedia of Biometrics, 2nd Edition (Accepted).

Journals
  • Biometric Quality: A Review of Fingerprint, Iris, and Face

    S. Bharadwaj, M. Vatsa, and R. Singh

    EURASIP Journal on Image and Video Processing, 2014.

     

  • Recognizing Surgically Altered Face Images using Multi-objective Evolutionary Algorithm

    H.S. Bhatt, S. Bharadwaj, R. Singh, and M. Vatsa

    IEEE Transactions on Information Forensics and Security (TIFS), 2013.

     

  • Memetically Optimized MCWLD for Matching Sketches with Digital Face Images

    H.S. Bhatt, S. Bharadwaj, R. Singh, and M. Vatsa

    IEEE Transactions on Information Forensics and Security (TIFS), 2012.

     

  • Plastic Surgery: A New Dimension to Face Recognition

    R. Singh, M. Vatsa, H.S. Bhatt, S. Bharadwaj, A. Noore and S.S. Nooreyezdan

    IEEE Transactions on Information Forensics and Security (TIFS), 2010.

Peer Reviewed Conferences